Center for Computational Imaging and Personalized Diagnostics (CCIPD) 2013 Annual Report Director: Anant Madabhushi, PhD Associate Professor, Department of Biomedical Engineering CCIPD Website: http://engineering.case.edu/centers/ccipd/ 12/31/2013 2 CCIPD in 2013 525 Wickenden Building Dept. of Biomedical Engineering Case Western Reserve University 2071 Martin Luther King Drive Cleveland, Ohio 44106-7207 Faculty Offices Center Space 3 Center Members Group dinner in New Jersey, Dec. 2013 Group Dinner in Cleveland 4 CCIPD Members Center Director: Anant Madabhushi, PhD Research Faculty Graduate Students Satish Viswanath, PhD Pallavi Tiwari, PhD Rachel Sparks (Rutgers, NJ) Ajay Basavanhally (Rutgers, NJ) George Lee (Rutgers, NJ) Rob Toth (Rutgers, NJ) Shoshana Ginsburg Sahir Ali Asha Singanamalli Prateek Prasanna Bharath Atthe Research Associates Mirabela Rusu, PhD Mahdi Orooji, PhD Haibo Wang, PhD Andrew Janowczyk (PhD) Scientific Programmer Ahmad Algohary Visiting Students Angel Cruz (Colombia) Geert Litjens (Nijmegen) Adminstrative Assitant Ann Tillet Undergraduate Students Eileen Huang (Rutgers, NJ) Aparna Kannan Michael Yim Gregory Penzias Srivathsan Babu Prabu 5 Center Members Center Director Anant Madabhushi, PhD Research Faculty Pallavi Tiwari, PhD Mahdi Orooji, PhD Ann Tillett Satish Viswanath, PhD Research Associates Mirabela Rusu, PhD Administrative Assistant Haibo Wang, PhD Scientific Programmer Andrew Janowczyk, PhD Ahmad Algohary 6 Center Members Graduate Students Sahir Ali Rachel Sparks Ajay Basavanhally Rob Toth George Lee Asha Singanamalli Prateek Prasanna Shoshana Ginsburg Bharath Atthe 7 Recent Alumni Tao Wang, PhD Raghav Padmanabhan, PhD Computational Biologist at GE Healthcare 8 Student Accomplishments Group photo with Ajay Basavanhally and George Lee immediately after their successful PhD thesis defenses, December2013 Group photo with Rachel Sparks and Robert Toth immediately after their successful PhD thesis defenses, December 2013 9 Conference Participation 2013 MICCAI in Japan: Angel Cruz EMBC in Japan: Eileen Huang 10 Conference Participation 2013 Mirabela Rusu and Rob Toth: “SeguiTM” algorithm wins first place in prostate zone segmentation challenge, ISBI 2013 George Lee and Mirabela Rusu: ISBI 2013 Mirabela Rusu and Pallavi Tiwari: Radiogenomics Meeting 2013 11 Summary of Accomplishments 2013 Center Members: 24 Faculty: 3 Research Associates: 4 Graduate Students: 9 Undergraduate Students: 5 Scientific Programmer: 1 Administrative Assistant: 1 Theses (5): 4 PhD + 1 MS Books: 1 Peer-Reviewed Journal Papers: 15 Peer-Reviewed Conference Papers: 32 Peer Reviewed Abstracts: 17 Issued Patents: 3 Patents pending: 1 Invention Disclosures: 6 Awarded Grants: 6 Awarded Fellowships: 2 Ongoing Projects: 40 12 Peer Reviewed Publications for 2013 Summary 36 32 28 24 20 16 12 8 4 0 Books Journal Papers Conference Papers Abstracts 13 Books Gurcan, M, Madabhushi, A, Medical Imaging 2013: Digital Pathology (Proceedings Volume), Proceedings of SPIE Volume: 8676, ISBN: 9780819494504, 2013. Peer Reviewed Publications Journal Papers Agner, S, Rosen, M, Englander, S, Thomas, K, Tomaszewski, J, Feldman, M, Zhang, P, Mies, C Schnall, M, Madabhushi, A, “Computerized Image Analysis for Identifying Triple Negative Breast Cancers and Distinguishing Triple Negative Breast Cancers from Other Molecular Subtypes of Breast Cancer on DCE-MRI: A Feasibility Study”, Radiology, Accepted. Toth, R.; Shih, N.; Tomaszewski, J.; Feldman, M.; Kutter, O.; Yu, D.; Paulus, J.; Paladini,G.; Madabhush, A.; "Histostitcher™: An Informatics Software Platform for Reconstructing Whole-Mount Prostate Histology using the Extensible Imaging Platform (XIP™ ) Framework“, Journal of Pathology Informatics (JPI), Accepted. Toth, R, Traughber, B, Ellis, R, Kurhanewicz, J, Madabhushi, A, “A Domain Constrained Deformable (DoCD) Model for Co-registration of Pre- and Post-Radiated Prostate MRI”, Neurocomputing, Special Issue on Image Guided Interventions, Accepted. Litjens, G, Toth, R, van de Ven, W, Hoeks, C, Kerkstra, S, van Ginneken, B, Reisaeter, L, Graham, V, Guillard, G, Birbeck, N, Zhang, J, Strand, R, Malmberg, F, Ou, Y, Davatzikos, C, Kirschner, M, Jung, F, Yuan, J, Qui, W, Gao, Q, Edwards, P, Maan, B, van der Heijden, F, Ghose, S, Mitra, J, Dowling, J, Barratt, D, Huisman, H, Madabhushi, A, “Evaluation of prostate segmentation algorithms for MRI: the PROMISE12 challenge”, Medical Image Analysis, Accepted. 14 Peer Reviewed Publications Journal Papers (Contd.) Wan, T, Bloch, BN, Shabbar, D, Madabhushi, A, “A Learning Based Fiducial-driven Registration Scheme for Evaluating Laser Ablation Changes in Neurological Disorders”, Neurocomputing, Special Issue on Image Guided Interventions, Accepted. Lewis, J, Ali, S, Luo, J, Thorstad, W, Madabhushi, A, “A Quantitative Histomorphometric Classifier (QuHbIC) Identifies Aggressive Versus Indolent p16 Positive Oropharyngeal Squamous Cell Carcinoma”, American Journal of Surgical Pathology, [Epub ahead of print] (PMID: 24145650). Sparks, R, Madabhushi, A, “Explicit Shape Descriptors: Novel Morphologic Features for Histopathology Classification Medical Image Analysis”, Medical Image Analysis, vol. 17(8), pp. 9971009, doi: 10.1016/j.media.2013.06.002. [Epub ahead of print] (PMID: 23850744). Sparks, R, Madabhushi, A, “Explicit Shape Descriptors: Novel Morphologic Features for Histopathology Classification Medical Image Analysis”, Medical Image Analysis, vol. 17(8), pp. 9971009, doi: 10.1016/j.media.2013.06.002. [Epub ahead of print] (PMID: 23850744). Sparks, R, Madabhushi, A, “Statistical Shape Model for Manifold Regularization: Gleason Grading of prostate histology”, Computer Vision and Image Understanding (Special Issue on Machine Learning in Medical Imaging), vol. 117(9), pp. 1138-1146, (PMID: 23888106). Toth, R., Ribault, J., Gentile, J.C., Sperling, D., Madabhushi, A. “Simultaneous Segmentation of Prostatic Zones Using Active Appearance Models with Multiple Coupled Levelsets,” Computer Vision and Image Understanding (CVIU) Special Issue on Shape Modeling, 2013, . doi: 10.1016/j.cviu.2012.11.013. Basavanhally, A, Ganesan, S, Shih, N, Feldman, M, Tomazewski, J, Madabhushi, A, “Multi-Field-ofView Framework for Distinguishing Tumor Grade in ER+ Breast Cancer from Entire Histopathology Slides”, Institute of Electrical and Electronics Engineers (IEEE) Transactions on Biomedical Engineering, [Epub ahead of print] (PMID: 23392336). 15 Peer Reviewed Publications Journal Papers (Contd.) Janowczyk, A, Chandran, S, Madabhushi, A, “Quantifying local heterogeneity via morphologic scale: Distinguishing tumoral regions from stromal regions”, Journal of Pathology Informatics, pp 4-8, 2013. Tiwari, P, Kurhanewicz, J, Madabhushi, A, “Multi-Kernel Graph Embedding for Detection, Gleason Grading of Prostate Cancer via MRI/MRS”, Medical Image Analysis, [Epub ahead of print] (PMID: 23294985). Agner, S, Xu, J, Madabhushi, A, “Spectral Embedding based Active Contour (SEAC) for Lesion Segmentation on Breast Dynamic Contrast Enhanced Magnetic Resonance Imaging”, Medical Physics, 2013, vol. 40(3):032305. doi: 10.1118/1.4790466. (PMID: 23464337) (Cover Article). Madabhushi, A, Viswanath, S, Lee, G, Tiwari, P, Medical Image Informatics for Personalized Medicine, Critical Values, pp. 30-32, July 2013 (Invited). 16 Peer Reviewed Publications Peer-reviewed Conference Papers Ginsburg, S, Madabhushi, A, “Variable Ranking in Kernel PCA: Applications to Cancer Characterization on Digitalized Histopathology”, In Proc of Medical Image Computing and Computer Assisted Interventions (MICCAI), vol. 2, pp. 238-45, 2013. Ali, S, Lewis, J, Madabhushi, A, “Spatially aware Cell Clusters Graphs: Predicting outcome in HPV associated oropharyngeal tumors”, In Proc of Medical Image Computing and Computer Assisted Interventions (MICCAI), vol. 1, pp. 412-419, 2013. Lee, G, Ali, S, Veltri, R, Epstein, J, Christudass, C, Madabhushi, A, “Cell Orientation Entropy (COrE): Predicting Biochemical Recurrence from Prostate Cancer Tissue Microarrays”, In Proc of Medical Image Computing and Computer Assisted Interventions (MICCAI), vol. 3, pp. 396-403, 2013. Cruz Roa, A, Osorio, FA, Madabhushi, A, Romero, E, “A deep architecture for image representation learning, visual interpretability and automatic basal-cell carcinoma tumor detection”, In Proc of Medical Image Computing and Computer Assisted Interventions (MICCAI), vol. 2, pp. 403-410, 2013. Huang, E, Rusu, M, Sparks, R, Toth, R, Shabbar, D, Madabhushi, A, “Anisotropic Smoothing Regularization (AnSR) in Thirion's Demons Registration Evaluates Brain MRI Tissue Changes Post-Laser Ablation”, Institute of Electrical and Electronics Engineers (IEEE) Engineering in Medicine and Biology Conference, pp. 4006-09, 2013 (PMID: 24110610). Prasanna, P, Jain, S, Bhagat, N, Madabhushi, A, “Decision Support System for Detection of Retinal Diseases Using Smartphones”, 7th International Conference on Pervasive Computing Technologies for Healthcare, pp. 176-79, 2013. Wang, H, Madabhushi, A, “Discriminatively weighted multi-scale local binary patterns: Applications in prostate cancer diagnosis on T2w MRI”, International Symposium on Biomedical Imaging 2013, Accepted. 17 Peer Reviewed Publications Peer-reviewed Conference Papers (Contd.) Lee, G, Sparks, R, Ali, S, Feldman, M, Master, S, Shih, N, Tomaszewski, J, Madabhushi, A, “Co-occurring Gland Tensors in Localized Cluster Graphs: Quantitative Histomorphometry for Predicting Biochemical Recurrence for Intermediate Grade Prostate Cancer”, International Symposium on Biomedical Imaging 2013, Accepted. Wan, T, Bloch, BN, Shabbar, D, Madabhushi, A, “Learning Based Fiducial Driven Registration (LeFiR): Evaluating Laser Ablation Changes For Neurological Applications”, International Symposium on Biomedical Imaging 2013, Accepted. Rusu, M, Kurhanewicz, J, Madabhushi, A, “A prostate MRI atlas of biochemical failure following radiotherapy“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Hwuang, E, Karthigeyan, S, Agner, S, Rusu, M, Sparks, R, Shih, N, Tomaszewski, J, Rosen, M, Feldman, M, Madabhushi, A, “Spectral Embedding-based Registration (SERg) for Aligning Multimodal Prostate Histology and MRI“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Orooji, M, Madabhushi, A, “Joint image segmentation and feature parameter estimation using expectation maximization: application to transrectal ultrasound prostate imaging”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Litjens, G, Huisman, H, Elliot, R, Shih, N, Feldman, M, Viswanath, S, Bomers, J, Madabhushi, A, “Distinguishing prostate cancer from benign confounders via a cascaded classifier on multi-parametric MRI”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Litjens, G, Huisman, H, Elliot, R, Shih, N, Feldman, M, Viswanath, S, Futterer, J, Bomers, J, Madabhushi, A, “Distinguishing benign confounding treatment changes following laser ablation therapy from residual prostate cancer on MRI”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. 18 Peer Reviewed Publications Peer-reviewed Conference Papers (Contd.) Ginsburg, S, Rusu, M, Kurhanewicz, J, Madabhushi, A, “Computer-extracted texture features on T2w MRI to predict biochemical recurrence following radiation therapy for prostate cancer“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Singanamalli, A, Wang, H, Lee, G, Shih, N, Ziober, A, Rosen, M, Master, S, Tomaszewski, J, Feldman, M, Madabhushi, A, “Supervised Multi-View Canonical Correlation Analysis: Fused Multimodal Prediction of Disease Prognosis“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Wang, H, Cruz-Roa, A, Basavanhally, A, Gilmore, H, Shih, N, Feldman, M, Tomaszewski, J, Gonzalez, F, Madabhushi, A, “Cascaded Ensemble of Convolutional Neural Networks and Handcrafted Features for Mitosis Detection“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Cruz-Roa, A, Basavanhally, A, Gonzalez, F, Gilmore, H, Feldman, M, Ganesan, S, Shih, N, Tomaszewski, J, Madabhushi, A, “Automatic detection of invasive ductal carcinoma in whole slide images with Convolution Neural Networks“, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Tiwari, P, Rogers, L, Wolansky, L, Madabhushi, A, “Differentiating recurrent glioblastoma multiforme from radiation induced effects via texture analysis on multi-parametric MRI”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Tiwari, P, Shabbar, D, Madabhushi, A, “Identifying MRI markers to evaluate early treatment related changes post laser ablation for cancer pain management”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. Rusu, M, Bloch BN, Jaffe C, Rofsky N, Genega E, Lenkinski R, Madabhushi A, “Statistical 3D Prostate Imaging Atlas Construction via Anatomically Constrained Registration”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. 19 Peer Reviewed Publications Peer-reviewed Conference Papers (Contd.) Rusu M, Wang H, Golden T, Gow A, Madabhushi A, “Multi-Scale, Multi-Modal Fusion of Histological and MRI Lung Volumes for Characterization of Airways”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Wan T, Bloch BN, Madabhushi A, “A Novel Point-based Nonrigid Image Registration Scheme Based on Learning Optimal Landmark Configurations”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Tiwari, P, Danish, S, Wong, S, Madabhushi A, “Quantitative Evaluation of Multi-parametric MR Imaging Marker Changes Post-laser Interstitial Ablation Therapy (LITT) for Epilepsy”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Singanamalli, A, Sparks, R, Rusu, M, Shih, N, Ziober, A, Tomaszewski, JE, Rosen, M, Feldman, M, Madabhushi, A “Correlating in vivo Imaging and ex vivo Vascular Markers to Identify Aggressive Prostate Cancer: Preliminary Results”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Prabu SB, Toth R, Madabhushi A, “A Statistical Deformation Model (SDM) based Regularizer for Non-rigid Image Registration: Application to registration of multimodal prostate MRI and histology”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Sparks R, Feleppa E, Barratt D, Bloch BN, Madabhushi A, “Fully Automated Prostate MRI and Transrectal Ultrasound (TRUS) Fusion via a Probabilistic Registration Metric”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Ginsburg SB, Bloch BN, Rofsky NM, Genega E, Lenkinski RN, Madabhushi A, “Iterative Multiple Reference Tissue Method for Estimating Pharmacokinetic Parameters on Prostate DCE MRI”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. 20 Peer Reviewed Publications Peer-reviewed Conference Papers (Contd.) Ali S, Veltri R, Epstein J, Christudass C, Madabhushi A, “Cell Cluster Graph for Prediction of Biochemical Recurrence in Prostate Cancer Patients from Tissue Microarrays”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Wang H, Rusu M, Golden T, Gow A, Madabhushi A, “Mouse Lung Volume Reconstruction from Efficient Groupwise Registration of Individual Histological Slices with Natural Gradient”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Basavanhally A, Madabhushi A, “EM-based Segmentation-Driven Color Standardization of Digitized Histopathology”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. Viswanath S, Sperling D, Lepor H, Futterer J, Madabhushi A, “Quantitative evaluation of treatment related changes on multi-parametric MRI after laser interstitial thermal therapy of prostate cancer, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2013. 21 Peer Reviewed Abstracts Danagoulian, J, Keither, JC, Hamilton, J, Hua, N, Pham, T, Madabhushi, A, Wan, T, Phinikaridou, A, Buckler, A, “Quantitative Assessment of vascular disease severity: Development of a Theranostic Imaging Biomarker”, Accelerating Development & Advancing Personalized Therapy (ADAPT) 2013, November 4-6, 2013, Cambridge, MA, Accepted (Poster). Madabhushi, A, Doyle, S, Basavanhally, A, Gilmore, H, Feldman, M, Mies, C, Tomaszewski, J, Harris, L, Ganesan, S, “Computer extracted image measurements of nuclear shape and texture from H&E images appear to stratify low and high risk ER+ breast cancers assessed via Oncotype DX“, 2013 San Antonio Breast Cancer Symposium, December 10- 14, 2013 in San Antonio, Texas, Accepted (Poster). Madabhushi, A, Basavanhally, A, Doyle, S, Wan, T, Singanamally, A, Thompson, C, Gilmore, H, Plecha, D, Harris, L, “Computer extracted image texture features on T2−weighted MRI appear to correlate with nuclear morphologic descriptors from H&E-stained histopathology in estrogen receptor positive breast cancers”, 2013 San Antonio Breast Cancer Symposium, December 10- 14, 2013 in San Antonio, Texas, Accepted (Poster). Madabhushi, A, Wan, T, Pletcha, D, Bloch, BN, Jaffe, C, Thompson, S, Gilmore, H, Harris, L, , “Computer derived image features on DCE-MRI appear to distinguish Estrogen Receptor-positive breast cancers with low and high Oncotype DX recurrence scores”, 2013 San Antonio Breast Cancer Symposium, December 10- 14, 2013 in San Antonio, Texas, Accepted (Poster). Tiwari, P, Danish, S, Madabhushi, A, “How long does the hippocampus take to settle down after MRIguided laser ablation for refractory epilepsy? Proof of concept using a multi-parametric analysis of MRI markers“, Annual Meeting of the American Epilepsy Society, 2013. Wan, T, Pletcha, D, Bloch, BN, Jaffe, C, Thompson, S, Gilmore, H, Harris, L, Madabhushi, A, “Computer Derived Texture Features on DCE-MRI Can Separate ER+ Breast Cancers with Low and High Oncotype DX Scores”, Proceedings of the Radiologic Society of North America 2013, Accepted. 12/31/2013 22 Peer Reviewed Abstracts (cont.) Ginsburg, S, Bloch, BN, Rofsky, NM, Genega, E, Lenkinski, RE, Madabhushi, A, “Tumors in the Peripheral Zone and Central Gland of the Prostate Have Different Perfusion Characteristics on Dynamic Contrast-enhanced MRI”, Proceedings of the Radiologic Society of North America 2013, Accepted. Tiwari, P, Rogers, L, Wolansky, L, Madabhushi, A, “Computerized image analysis of texture descriptors in multi-parametric MRI to distinguish recurrent glioblastoma multiforme from radiation necrosis”, 4th Quadrennial Meeting of the World Federation of Neuro-Oncology held in conjunction with the 2013 Scientific Meeting and Education Day of the Society for Neuro-Oncology, San Francisco, CA, November 2013. Viswanath, S, Madabhushi, A, “Bias field correction and intensity standardization improve the diagnostic accuracy of multi-parametric MRI in computerized detection of prostate cancer in vivo”, Society for Imaging Informatics in Medicine (SIIM), Dallas, TX, 2013 Buckler, A, Keith, J, Madabhushi, A, Hamilton, J, “Non-invasive theranostic to predict and assess response to atherosclerotic drugs”, Experimental Biology, 2013. Toth, R, Kurhanewicz, J, Madabhushi, A, “Registration of Pre and Post Intensity Modulated Radiation Therapy Prostate MRI for Quantification of MR Imaging Marker Changes and Precise Local Prostate Deformations”, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. Singanamally, A, Sparks, R, Rusu, M, Shih, N, Ziober, A, Tomaszewski, J, Rosen, M, Feldman, M, Madabhushi, A, “Radiomics Driven Image Analysis and Co-registration Scheme to Identify DCE MRI Markers for Microvascular Density“, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. Tiwari, P, Shabbar, D, Wong, S, Madabhushi, A, “Quantitative Study of changes in multi-parametric MRI markers post-laser interstitial ablation therapy (LITT) for epilespy“, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. 12/31/2013 23 Peer Reviewed Abstracts (cont.) Tiwari, P, Kurhanewicz, J, Madabhushi, A, “A quantitative framework to study MRI related treatment changes in the prostate post-IMRT“, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. Viswanath, S, Futterer, J, Lepor, H, Sperling, D, Madabhushi, A, “Quantitative Evaluation of Treatment Related Changes on Multi-Parametric MRI after Laser Interstitial Thermal Therapy of Prostate Cancer “, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. Wan, T, Wong, S, Shabbar, D, Madabhushi, A, “Co-registration of MRI via a Learning Based Fiducialdriven Registration (LeFiR) Scheme: Evaluating Laser Irradiation Changes for Glioblastomas and Epilepsy“, The International Society for Magnetic Resonance in Medicine (ISMRM), 2013. Lewis, J, Ali, S, Thorstad, W, Madabhushi, A, “A Quantitative Histomorphometric Classifier Identifies Aggressive Versus Indolent p16 Positive Oropharyngeal Squamous Cell Carcinoma”, United States and Canadian Academy of Pathology's 102nd Annual Meeting, 2013, Accepted. 12/31/2013 24 Invited Lectures “Image Analytics in Digital Pathology: Prognosis prediction and data enrichment”, Department of Ophthalmology, Case Western Reserve University, Cleveland, OH, November 8th, 2013. “Image Analytics in Digital Pathology: Prognosis prediction and data enrichment”, Plenary Lecture, First Nordic Digital Pathology Conference, Linkoping, Sweden, October 31, 2013. “Computer aided diagnosis: Nuts and Bolts”, Department of Radiology, Case Western Reserve University, Cleveland, OH, October 22, 2013. “Convergence of multi-scale, multi-modal data: How to get more out of radiologic imaging via pathology, omics, and image analytics”, Grand Rounds, Department of Radiology, Case Western Reserve University, Cleveland, OH, October 22, 2013. “Computerized Image Based Predictor of Aggressiveness in ER+ Breast Cancers”, Imaging Center Seminar Series, Case Western Reserve University, September 16th, 2013. “Image Analytics and Radiomics”, Workshop on Imaging and Genomics, National Cancer Institute, Bethesda, MD, June 27th, 2013. “Image Analytics and Personalized Medicine in Prostate Cancer: Who to treat? How to Treat? Where to treat? Did it work?”, Department of Urology, Weil Cornell Medical Center, New York, NY, June 17th, 2013. “Fusion of multi-scale, multi-modal data”, Ultrasonic Imaging and Tissue Characterization Symposium, Arlington, VA, June 12, 2013. “Image Analytics in Personalized Medicine: Who to treat? How to treat? Where to treat? Did it work?”, Grand Rounds in Radiation Oncology, Case Western Reserve University, May 21, 2013. “Quantitative data convergence: Correlating multi-scale, multi-modal data for disease diagnosis and prognosis”, Invited Seminar, Institute of Information Sciences in Imaging (ISIS), Stanford University, Palo Alto, March 20, 2013. 12/31/2013 25 Invited Lectures “Quantitative Image Biomarkers for MRI and histopathology: Applications in Breast Cancer Prognosis”, Breast Cancer Retreat, Case Comprehensive Cancer Center, Cleveland, OH, March 11, 2013. “Image based histologic risk score (IbRiS): Predicting Disease Aggressiveness and Patient outcome”, Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH, March 8, 2013. “Quantitative Histologic based Image Classifier (QuHbIC): Predicting Disease Aggressiveness and Patient outcome”, Grand Rounds in Pathology, Case Western Reserve University, February 27, 2013. “Quantitative Data Convergence: Use case in Prostate Cancer Imaging”, Invited Lecture, International Society for Magnetic Resonance in Medicine (ISMRM) workshop on MRI for Cancer Gone Multimodal, Valencia, Spain, February 19th-22nd, 2013. “Quantitative Histomorphometry and Radiogenomics: Applications to breast cancer prognosis”, Breast Cancer Research Incubator Meeting, Case Western Reserve University, January 17th, 2013. 12/31/2013 26 Patents Issued Patents “Malignancy Diagnosis Using Content-Based Image Retrieval of Tissue Histopathology”, Anant Madabhushi, Michael D. Feldman, John Tomaszewski, Scott Doyle, United States Serial Number (USSN): 8,280,132. "Computer Assisted Diagnosis (CAD) of cancer using Multi-Functional Multi-Modal in vivo Magnetic Resonance Spectroscopy (MRS) and Imaging (MRI)", Anant Madabhushi, Satish Viswanath, Pallavi Tiwari, Robert Toth, Mark Rosen, John Tomaszewski, Michael D. Feldman, United States Serial Number (USSN): 8,295,575. "Combined feature ensemble mutual information image registration", Anant Madabhushi, Jonathan Chappelow, Mark Rosen, John Tomaszewski, Michael Feldman, United States Serial Number (USSN): 8,442,285. Patents Pending “Classification of Biological Tissue by Multi-mode data registration, segmentation, and characterization”, Andrew Buckler, Anant Madabhushi, James Hamilton, Shannon Agner, Mark Rosen, US 2013/0202173 A1 27 Patents Invention Disclosures (cont.) “A Texture Based Finite Element Model Registration Scheme For Registering Pre-Treatment Intensity Modulated Radiation Therapy MR Imagery To Post-Treatment MR Imagery”, Anant Madabhushi, Robert Toth, Case 2013-2375. “Discriminatively weighted multi-scale local binary patterns for object detection”, Anant Madabhushi, Haibo Wang, Case 2013-2391 “Co-occurring Gland Tensors in Localized Cluster Graphs for Quantitative Histomorphometry”, Anant Madabhushi, George Lee, Sahir Ali, Rachel Sparks, Case 2013-2369. “Co-occurring Nuclear Tensors in Localized Cluster Graphs for Quantitative Histomorphometry”, Anant Madabhushi, George Lee, Sahir Ali, Rachel Sparks, Case No. 2013-2452. “Spatially aware Cell Cluster (SpACCl) Graphs for Quantitative Histomorphometry”, Anant Madabhushi, Sahirzeeshan Ali, Case No. 2013-2471. “Cascaded Ensemble of Convolutional Neural Networks and Handcrafted Features For Breast Cancer Diagnosis”, Anant Madabhushi, Haibo Wang, Angel Cruz, Case No. 2014-2572. 28 Awards and Accomplishments in 2013 Honors, Awards and Recognition First place Scientific award for “Bias field correction and intensity standardization improve the diagnostic accuracy of multi-parametric MRI in computerized detection of prostate cancer in vivo” at the Society for Imaging Informatics in Medicine (SIIM) meeting, Dallas, TX, June 7th, 2013 First Place in “National Cancer Institute International Society for Biomedical Imaging (NCI-ISBI) 2013 Challenge - Automated Segmentation of Prostate Structures” held in conjunction with Institute of Electrical and Electronics Engineers (IEEE) International Symposium on Biomedical Imaging, San Francisco, CA, April 7th, 2013. Paper “Spectral embedding based active contour (SEAC) for lesion segmentation on breast dynamic contrast enhanced magnetic resonance imaging” featured on cover of Medical Physics, April 2013. Media Recognition “Teaching Computers to Tell Cancer Cells Apart”, Prostate Cancer Discovery, A Publication of the Patrick C. Walsh Prostate Cancer Research Fund, vol. 10, Winter 2014. “Digital Pathology can solve crisis”, Linkoping University, News, Faculty of Health Sciences, Nov. 2013. “New Center Makes Diagnoses more personal”, Case School of Engineering, Annual Report 2012-2013, pp. 11, Nov. 2013. “Biomedical engineering’s Anant Madabhushi to serve on editorial board for new Journal on Medical Imaging”, The Daily, September 27th, 2013. “Battling Brain Radiation Necrosis”, Innovations in Neurosciences, pp. 3, Fall 2013. “Team from the Center for Computational Imaging and Personalized Diagnostics Receives Coulter Translational Award”, Featured in Case Comprehensive Cancer Center Newsletter, August 5, 2013. 29 Awards and Accomplishments in 2013 Media Recognition (cont.) “Dr. Madabhushi awarded breast cancer grant from Ohio Third Frontier”, Featured on: • The Daily, July 8th, 2013. • Case Comprehensive Cancer Center Newsletter, June 24th, 2013. “Biomedical Engineering’s faculty members receive scientific award”, The Daily, June 14th, 2013. “Anant Madabhushi’s biomedical engineering team awarded multiple patents; Madabhushi invited to conference”, The Daily, June 13th, 2013. “Using big data to identify prostate cancers and vest treatments”, Press Release, May 23rd, 2013. Featured on: • Phys.org • Eurekalert.org • Freenewspos.com • News-medical.net • www.healthcanal.com • Bio-medicine.org • Mdxlinx.com/urology • Topnews.us • Textradar.com • primeurmagazine.com • freenewspos.com “Biomedical engineering research lab aims to improve cancer treatments”, Feature Story, The Daily, May 22nd, 2013. “Biomedical engineering’s Anant Madabhushi awarded clinical trial grant”, The Daily, May 17th, 2013. 30 Awards and Accomplishments in 2013 Media Recognition (cont.) “BME Student Wins Second Place in Imaging Conference”, Featured in: • Case Comprehensive Cancer Center Newsletter, May 6th, 2013. • The Daily, May 9th, 2013. “CWRU team takes first at National Cancer Institute Competition”, Featured in: • Case Comprehensive Cancer Center Newsletter, April 15th, 2013. • The Daily, April 15th, 2013. “Biomedical engineering faculty member’s article featured on cover of “Medical Physics”, Featured in The Daily, April 5, 2013. “Technologies to aid stroke and Prostate Cancer Patients win grants from Philadelphia Program”, TechLifeSciNews, pp. 23, March 2013. “Strong Cancer related funding week for Anant Madabhushi and Team”, Case Comprehensive Cancer Center Newsletter, Jan 14th, 2013. “vascuVis awarded NSF SBIR Phase I funding”, Spin-Off and Start-up News, January, 2013. 31 Professional/Editorial Activities in 2013 Chairing, Membership Program Committees of Conferences, Workshops, Special issues IAPR TC20 technical committee member for Pattern Recognition for Bioinformatics, 2014. Session Chair, Conference 9401: Digital Pathology, Keynote Session, International Society for Optics and Photonics (SPIE) Medical Imaging, Feb 18th, 2014, San Diego, CA. Co-organizer and Co-Chair, Workshop: What do pathologists see on a slide? Implications for digital pathology, International Society for Optics and Photonics (SPIE) Medical Imaging, Feb 18th, 2014, San Diego, CA. Scientific Selection Committee Member, Special issue of Computerized Medical Imaging and Graphics, Elsevier on "Breakthrough technologies in digital histopathology", 2013. Chair, Oral Session “Microscopy, histology, and computer aided diagnosis”, Medical Image Computing and Computer Assisted Intervention Society (MICCAI) 2013, September 24th, 2012, Nagoya, Japan. IAPR TC20 technical committee member for Pattern Recognition for Bioinformatics, July 2013. Participant and Panelist, National Cancer Institute’s Cancer Imaging Program (CIP) sponsored workshop, “Correlating Imaging Phenotypes with Genomic Signatures”, NIH Campus-Natcher, June 26-27th, 2013. Co-organizer and Co-Chair, Fourth workshop on Content-based retrieval for clinical decision support, Medical Image Computing and Computer Assisted Intervention Society (MICCAI) 2013, Nagoya, Japan. Co-Organizer and Co-Chair, Grand Challenge on Segmentation of Prostatic Structures from MRI, Medical Image Computing and Computer Assisted Intervention Society (MICCAI) 2013, Nagoya, Japan. NIH Clinical Micro-dissection Working Group, 2013-Present 32 New Grants Awarded in 2013 Madabhushi, Anant (PI) 09/01/13-08/31/16 DOD Prostate Cancer Synergistic Idea Development Award (PC120857) $1,178,117/Award = $570,618 Active surveillance biomarkers for reclassification of Small Low Grade, Low Stage T1c Indolent Prostate Cancer Tumors Role: Partnering PI with Robert Veltri the initiating PI Gulani, Vikas (PI) 06/01/13 - 05/30/18 NIH, R01DK098503-02 Comprehensive Quantitative Ultrafast 3D Liver MRI Role: Co-I (Madabhushi) Buckler, Andrew (PI), 01/01/13 - 06/01/13 NSF IIP-1248316, Computer assisted prognosis of debilitating disease Role: Co-I (Madabhushi) Madabhushi, Anant (PI) 09/01/13-08/31/14 Ohio Third Frontier Program Histologic Image based predictors for treatment prediction in ER+ breast cancers Role: PI Tiwari, Pallavi (PI) 09/01/13 - 08/31/14 Coulter Research Translational Partnership NeuroRadVisionTM: Image based risk score prediction of recurrent brain tumors 33 Grants and Research Supports in 2013 Competitive Fellowship-Sponsored Postdoctoral Advisees – Anant Madabhushi Rusu, Mirabela (PI) 04/01/13 - 03/30/13 Post-doctoral Training Fellowship $125,500 Department of Defense “Identifying MRI markers for viable, post-radiation prostate tumors in men undergoing radiation therapy followed by prostatectomy” Role: Mentor 34 Student, Post-doctoral Awards and Fellowships Eileen Hwuang, SPIE Medical Imaging Student Grant ($200), 2013 Asha Singanamally, SPIE Medical Imaging Student Grant ($200), 2013 Angel Alfonso Cruz Roa, Medical Image Computing and Computer Assisted Intervention Society (MICCAI) Travel Award ($500), 2013. Mirabela Rusu, Postdoc representative for Case Western Reserve University Convocation, 2013. Mirabela Rusu, First Place in Postdoctoral Poster Presentation Award, Research ShowCASE ($550), 2013. George Lee, Travel award to attend Intl Symposium on Biomedical Imaging ($350), 2013. Haibo Wang, Tutorial award to attend Intl Symposium on Biomedical Imaging ($150), 2013. Eileen Hwang, Scholarship to BME REU Program at Case Western Reserve University ($4900), 2013. Srivatsan Babu Prabhu, Scholarship to BME REU Program at California Institute of Technology ($4900), 2013. Shannon Agner, Outstanding MD/PhD Student Award, UMDNJ/Rutgers, March 15, 2013.Eileen Hwuang, Research grant ($626), Institute of Electrical and Electronics Engineers (IEEE) Princeton/Central Jersey Chapter, 2013Asha Singanamally, Michael B. Merickel Best Student Paper Award, Runner Up ($500), SPIE Medical Imaging, 2013 Rachel Sparks, Graduate School of New Brunswick Travel Award ($200), 2013 35 ACTIVE RESEARCH METHOD S MULTIMODAL DATA INTEGRATION COMPUTER AIDED DIAGNOSIS IMAGE SEGMENTATION MACHINE LEARNING IMAGE REGISTRATION RADIOLOGY HISTOPATHOLOGY BIOINFORMATICS PROSTATE CANCER BREAST CANCER MEDULLOBLASTOMA OROPHARYNGEAL CANCER LUNG CANCER 36 DIMENSIONALITY REDUCTION MULTI-MODAL REGISTRATION FEATURE SELECTION CLASSIFICATION METHODS FEATURE EXTRACTION IMAGE SEGMENTATION ACTIVE LEARNING & CONTENT BASED IMAGE RETRIEVAL KNOWLEDGE REPRESENTATION AND MULTI-MODAL DATA FUSION IMAGE REGISTRATION AND SEGMENTATION 38 Multi-Attribute Probabilistic Prostate Elastic Registration (MAPPER) R. Sparks, B. N. Bloch, E. Feleppa, D. Baratt, A. Madabhushi. Fully automated prostate magnetic resonance imaging and transrectal ultrasound fusion via a probabilistic registration metric. In Proc. SPIE 86710A, 2013. 39 Spectral Embedding-Based Registration (SERg) Aligning Multimodal Prostate Histology and MRI Performing registration on spectral embedding representations of multimodal images, i.e. parametric eigenvector images (PrEIms), results in alignment superior to that of direct registration of grayscale images when using intensity difference-based similarity metrics. Demonstration SERg on synthetic brain images Intensity MRI and histology images MRI and histology PrEIms Demons Demons registration registration performed on PrEIms E. Hwuang, S. Karthigeyan, S. C. Agner, M. Rusu, R. Sparks, N. Shih, J. E. Tomaszewski, M. Rosen, M. Feldman, and A. Madabhushi. “Spectral Embedding-Based Registration (SERg) Aligning Multimodal Prostate Histology and MRI.” Accepted SPIE Medical Imaging 2014. 40 Prostatome: A Combined Anatomical and Disease Based MRI Atlas of the Prostate Top: Statistical shape atlas of the central gland, peripheral zone and prostate Middle: spatial distribution of cancer relative to the prostate statistical shape Bottom: spatial distribution of cancer relative to the central gland and peripheral zone shape Rusu et. al. Prostatome: A combined anatomical and disease based MRI atlas of the prostate, Medical Physics (In Review) 41 Simultaneous Segmentation of Prostatic Zones via Active Appearance Models with Multiple Coupled Level-sets T2-weighted MRI Prostate (yellow) is comprised of central zone (red) and peripheral zone (purple) Toth, R., Ribault, J., Gentile, J.C., Sperling, D., Madabhushi, A. “Simultaneous Segmentation of Prostatic Zones Using Active Appearance Models with Multiple Coupled Levelsets,” CVIU Special Issue on Shape Modeling 117(9), Sep 2013. pp. 1051-1060. doi: 10.1016/j.cviu.2012.11.013. 42 Domain-Constrained Deformable (DoCD) Model Biomechanical finite element model (FEM) for spatially registering pre-, post-treatment prostate MRI FEM Explicit shrinking due to radiation treatment modeled Necessary for determining voxellevel treatment changes RMS error of manually identified internal prostate fiducials <3mm R. Toth, B. Traughber, R. Ellis, J. Kurhanewicz, and A. Madabhushi. Registering pre-, post-radiation treatment prostate imagery using domain-specific biomechanical model. Neurocomputing, Special Issue on Image Guided Interventions: Accepted, 2013. 43 43 Fully Automatic Unsupervised Transrectal Ultrasound Image Segmentation and Parameter Estimation Results of Segmentation Results of Parameter Estimation ˆ argmax Zˆ , P (D | Z, ) Z , 0.04 0.035 0.025 0.03 Probability Probability 0.02 0.025 0.02 0.015 0.015 0.01 0.01 0.005 0.005 0 0 0.5 1 1.5 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Cramer Rao Lower Band Orooji, M, Madabhushi, A, “Joint image segmentation and feature parameter estimation using expectation maximization: application to transrectal ultrasound prostate imaging”, The International Society for Optics and Photonics (SPIE) Medical Imaging, 2014. 44 FEATURE EXTRACTION 45 Explicit Shape Descriptors for Cancer Aggressiveness 1.Shape Model Construction 3. Shape Dissimilarity 2. Shape Model Registration Gleason pattern 3 from Gleason pattern 4 Sensitivity Sensitivity Benign from Malignant 4. Dimensionality Reduction 1-Specificity 1-Specificity R. Sparks and A. Madabhushi. Explicit shape descriptors: novel morphologic features for histopathology classification. Medial Image Analysis 17: 997-1009, 2013. 46 Pharmacokinetic Modeling for Prostate Cancer Localization • Iterative multiple reference tissue method (IMRTM) for pharmacokinetic (PK) feature extraction from DCE MRI – Patient-specific values for Ktrans and ve are iteratively estimated for voxels in tissues A and B. – No need for arterial input function – Does not rely on population-based PK parameters • Preliminary results shown on the right for a CG tumor Ground truth Ktrans ve Ginsburg S., et al. Tumors in the peripheral zone and central gland of the prostate have different perfusion characteristics on dynamic contrast-enhanced MRI. RSNA 2013. 47 Cell Orientation Entropy (COrE) Features Stratify More and Less Aggressive Prostate Cancer on Tissue Microarrays Aggressive cancer (left) shows more disorder in orientation of the nuclei compared to less aggressive cancer (right) Lee, G, Ali, S, et al., “Cell Orientation Entropy (COrE): Predicting Biochemical Recurrence from Prostate Cancer Tissue Microarrays”, In Proc of Medical Image Computing and Computer Assisted Interventions (MICCAI), vol. 3, pp. 396-403, 2013. 48 Co-Occurring Gland Tensors for Differentiating Between Benign and Malignant Glands in Prostate Cancer • • Co-occurrence Matrices reveal the difference between malignant and benign glands in CaP tissue Brighter pixels in the off-diagonal components signify frequent co-occurrence of glands of different directionality AUC = 0.7959 Malignant prostate glands Brighter off-diagonal components Dimmer off-diagonal Normal benign components prostate glands Lee, G, Sparks, R, Ali, S, Feldman, M, Master, S, Shih, N, Tomaszewski, J, Madabhushi, A, “Co-occurring Gland Tensors in Localized Cluster Graphs: Quantitative Histomorphometry for Predicting Biochemical Recurrence for Intermediate Grade Prostate Cancer”, International Symposium on Biomedical Imaging 2013 49 Spatially Aware Cell Cluster (SpACCl) Graphs: Predicting Outcome in Oropharyngeal p16+ Tumors Classify location of nuclei as stroma or epithelium via Superpixel Build separate graph for each tissue type Extract separate features from each tissue class Representative TMA spot images of H&E stained histology for (a) non-progressing and (d) progressing p16+ OSCC with (b), (e) associated cell cluster identification with epithelium nodes in yellow, and stromal nuclei in green and (c), (f) cell cluster graphs. Note differences in CCGs for the two classes of OSC (top versus bottom row). S Ali, J Lewis, A Madabhushi, Spatially Aware Cell Cluster (SpACCl) Graphs: Predicting Outcome in Oropharyngeal p16+ Tumors - Medical Image Computing and Computer-Assisted Interventions, 2013 50 MACHINE LEARNING 51 Variable Importance in Nonlinear Kernels for Feature Reduction in High-Dimensional Classification Problems Step 1: Feature Extraction Step 2: Dimensionality Reduction Prostate cancer: biochemical recurrence Features selected by VINK, in conjunction with logistic regression classifiers, lead to similar areas under the ROC curves to the entire high-dimensional feature set: applications in (a) prostate cancer, (b) breast cancer, and (c) oropharyngeal cancer. Step 3: Assess Variable Importance 1 0.95 Prostate cancer: no biochemical recurrence 0.9 0.85 AUC 0.8 0.75 0.7 Breast cancer: low Oncotype DX score 0.6 0.55 Laplacian eigenmap Gaussian KPCA Isomap 0.5 1.1 All features PCA Gaussian KPCA Top 5 features Isomap Laplacian Eigenmap 1 0.9 Step 3: Compute VINK AUC Breast cancer: high Oncotype DX score 0.65 0.8 0.7 Oropharyngeal cancer: progressor Assess variable contributions to classification on the embedding 0.6 0.5 All features Top 2 features All features Top 5 features 160 150 140 130 120 MSE Oropharyngeal cancer: Non-progressor 110 100 90 80 Ginsburg S, “Variable importance in nonlinear kernels (VINK): Classification of digitized histopathology,” In Proc. MICCAI 2013. 70 60 52 Supervised Multiview CCA: Data Fusion Strategy Goal: Identification of ex vivo markers of 5 year prostate cancer recurrence Representations of Histology Proteomic Expression Singanamalli et al. “Supervised Multi-View Canonical Correlation Analysis: Fused Multimodal Predictors of Disease Prognosis”. To appear in proc SPIE 2014. 53 Manifold Regularization via Statistical Shape Model of Manifolds ℳ SSMM Classification results for ℳ and SSMM on prostate histopathology. Identifies noisy samples on ℳ which cause classification errors. Removal of noisy samples results in a manifold SSMM, with better defined decision boundaries R. Sparks and A. Madabhushi. Statistical shape model for manifold regularization: Gleason grading of prostate histology. Computer Vision and Image Understanding 117: 1138-1146, 2013. 54 Predicting Error Rates for Large Datasets From Smaller Cohorts Goal: Capture the variability of small dataset lost during random repeated sampling Solution: Rotate training and testing pools using K-fold cross-validation sampling First, a dataset is partitioned into training and testing pools using a K-fold sampling strategy. Each of the K training pools undergoes repeated random sampling (RRS), in which error rates are calculated at different training set sizes via a subsampling procedure. A permutation test is used to identify statistically significant error rates, which are then used to extrapolate learning curves and predict error rates for larger datasets. 55 TREATMENT EVALUATION 56 A Learning Based Fiducial-driven Registration Scheme for Evaluating Laser Ablation Changes in Neurological Disorders (A) (A) Shows the identification of important landmarks using the deformed synthetic images. (B) illustrate the prediction of optimal configuration of landmarks for new images. The estimated landmark set can be applied to real clinical deformation to recover the focal deformation. (1) (2) (3) (4) (5) (6) (7) (8) (B) The first and second rows show two different registration experiments performed on an Epilepsy case and a glioblastoma multiforme (GBM) case. (1),(2) and (5),(6) are pre- and post- laser induced interstitial thermal therapy (LITT) images. (3),(4) and (7),(8) are the different maps between the registered and pre-LITT images using the presented LeFiR method and a uniform landmark spacing method. T.Wan, B.N.Bloch, S.Danish, A.Madabhushi, A Learning Based Fiducial-driven Registration Scheme for Evaluating Laser Ablation Changes in Neurological Disorders, accepted by Neurocomputing, 2013. 57 A Preliminary Quantitative Study of Changes in MultiParametric MRI Markers Post Ablation for Epilepsy Tiwari et al., ISMRM 2013, SPIE 2013. 58 Identifying MRI Markers for Treatment Changes Post-Laser Ablation for Cancer Pain Management Tiwari et al., SPIE 2014. 59 Anisotropic Smoothing Regularization (AnSR) in Thirion’s Demons Registration Evaluates Brain MRI Tissue Changes Post-Laser Ablation Normalized sum of squared difference alignment error: Anisotropic smoothing regularizer (AnSR) utilizes edge-detection and denoising within the Demons registration framework to regularize the deformation field more aggressively in regions of homogeneously oriented displacements while simultaneously regularizing less aggressively in areas containing heterogeneous local deformation and tissue interfaces. This method contributes to assessment of shortterm treatment via laser-induced interstitial thermal therapy (LITT) by evaluating changes in posttreatment MRI as a measure of response. Hwuang, E, Danish S, Rusu M, Sparks R, Toth R, Madabhushi A. 2013. “Anisotropic smoothing regularization (AnSR) in Thirion's Demons registration evaluates brain MRI tissue changes post-laser ablation.” Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2013:4006-9. 60 Identifying Quantitative in Vivo Multi-Parametric MRI Features for Treatment Related Changes After LITT of Prostate Cancer Identifying most effective quantitative features from in vivo MP-MRI that can depict changes within the ablation zone as well as LITT-related effects. Red outline – CaP, green outline – normal LITT-related benign change Viswanath S, etal. “Identifying Quantitative In Vivo Multi-Parametric MRI Features For Treatment Related Changes after LITT of Prostate Cancer”, Neurocomputing, Under revision. 61 A Prostate MRI Atlas of Biochemical Failures Following Radiotherapy Per voxel difference between Biochemical and non biochemical recurrence MRI atlases are built for the patients with Biochemical Recurrence (1a) and population without biochemical recurrence (1b); The two atlases are compared on a per voxel basis as well as (3) morphologically. Morphological differences Rusu et. al. SPIE Medical Imaging 2014 (accepted). 62 COMPUTER AIDED DIAGNOSIS AND PROGNOSIS 63 Decision Support System for Distinguishing Recurrent Brain Tumors from Radiation Necrosis Computer-extracted texture descriptors distinguish recurrent brain tumors from treatmentinduced effects Recurrent Brain Tumor Radiation Necrosis Tiwari P, Prasanna P, Rogers L, etal., “Computerized image analysis of texture descriptors in multi-parametric MRI to distinguish recurrent brain tumor from radiation necrosis,” 18th Annual Society of Neuro-Oncology Scientififc Meeting, 2013. 64 Discriminatively Weighted Multi-Scale Binary Patterns for Prostate Cancer Detection on T2w MRI Multi-scale Local binary patterns Weight learning for each scale Classification via feature matching Figure. First row: groundtruth; Middle row: DWLBP; Bottom row: LBP Wang H, et al. Discriminatively weighted multi-scale local binary patterns: Applications in prostate cancer diagnosis on T2w MRI, ISBI 2013. 65 Evaluating the Need for Intensity Artifact Correction in MPMRI for Computerized Detection of Prostate Cancer In Vivo Demonstrated that explicit correction of intensity inhomogeneity and drift in all 3 MP-MRI protocols (T2w, DCE, DWI) prior to combining them within a classification scheme will yield improved CaP detection accuracy in vivo (won best scientific poster award) “Evaluating the need for intensity artifact correction in MP-MRI for computerized detection of prostate cancer in vivo”, SIIM, 2013 66 Radiohistomorphometry: Identification of Quantitative Radiological Imaging Biomarkers of Prostate Cancer Quantitative representation of tumor microvessel arrangement Correlation heatmap of DCE MRI and microvessel features Quantitative representation of tumor on corresponding in vivo DCE MRI Singanamalli et al. “Identifying in vivo DCE MRI parameters correlated with ex vivo quantitative microvessel architecture: A Radiohistomorphometric Approach”. In Proc SPIE 2013. 67 Correlation between Quantitative Imaging Biomarkers on T2wMRI and H&E Stained Histopathology in ER+ Breast Cancer Goal: Provide histological context to MRI-based QIBs by identifying highly correlated pairs of QIBs between the two modalities Histology: More than 2000 QIBs describing nuclear morphology & texture MRI: More than 650 QIBs describing lesion texture Madabhushi, A, Basavanhally, A, Doyle, S, Wan, T, Singanamally, A, Thompson, C, Gilmore, H, Plecha, D, Harris, L, “Computer extracted image texture features on T2−weighted MRI appear to correlate with nuclear morphologic descriptors from H&E-stained histopathology in estrogen receptor positive breast cancers”, 2013 San Antonio Breast Cancer Symposium, December 10- 14, 2013 in San Antonio, Texas 68 Identifying Molecular Subtypes of Breast Cancers on MRI via Texture Kinetics Agner, S, Rosen, M, Englander, S, Thomas, K, Tomaszewski, J, Feldman, M, Zhang, P, Mies, C Schnall, M, Madabhushi, A, “Computerized Image Analysis for Identifying Triple Negative Breast Cancers and Distinguishing Triple Negative Breast Cancers from Other Molecular Subtypes of Breast Cancer on DCE-MRI: A Feasibility Study”, Radiology, Accepted 69 Predicting Biochemical Recurrence Following Radiation Therapy for Prostate Cancer on Pre-Treatment T2w MRI Three texture features extracted from cancerous regions on pre-treatment T2w MRI can be used to discriminate between patients who will develop biochemical recurrence and those who will remain recurrence-free (below). KaplanMeier curves show improved prediction of recurrence-free survival for texture features (above, right) than for the Kattan nomogram (below, right). Ginsburg S, et al. Computer extracted texture features on T2w MRI to predict biochemical recurrence following radiation therapy for prostate cancer. SPIE 2014. 70 A Deep Learning Architecture for Image Representation, Visual Interpretability and Automated Basal-Cell Carcinoma Detection Goal: Basal Cell Carcinoma (BCC) detection (cancerous or healthy tissue) with good classification performance and visual interpretability (a) An unified deep learning method for image representation learning, classification and prediction interpretability was proposed for BCC detection. A database of squared histopathology image are used (1) to train a Unsupervised Feature Learning method and learn a set of visual features. (2) Then a convolutional auto-encoder and pooling layer is used to represent the histopathology images and (3) classify them between cancerours or healthy tissue with a soft-max classifier. Finally, (4) a novel strategy exploits learned representation to produce visual interpretable predictions like a digital stain. (b) Evaluation results show the good classification performance of method, its prediction and salient discriminatn maps (red:cancer, blue: healthy). a) b) Angel Cruz-Roa, John Arevalo, Anant Madabhushi, Fabio González. A Deep Learning Architecture for Image Representation, Visual Interpretability and Automated BasalCell Carcinoma Cancer Detection. MICCAI 2013, the 16th International Conference on Medical Image Computing and Computer Assisted Intervention. Nagoya (Japan), September 22-26 2013. Lectures Notes in Computer Sciences. Vol. 8150. pp 403-410. 12/06/13 71 Automatic Detection of Invasive Ductal Carcinoma in Whole Slide Images with Convolutional Neural Networks Goal: Invasive ductal carcinoma detection automatically in whole-slide image tissues of BCa. (1) Square image patches are sampled from tissue region of whole-slide images (WSI) (2) to train a 3-layer Convolutional Neural Network model (CNN). (3) The trained CNN model is posteriorly used to predict the probability of each square image patch sampled belongs to Invasive Ductal Carcinoma from a new WSI of Breast Cancer (BCa) to build a IDC probability map. Manual annotation Automatic annotation Manual annotation Automatic annotation Angel Cruz-Roa, Ajay Basavanhally, Fabio Gonzalez, Hannah Gilmore, Michael Feldman, Shridar Ganesan, Natalie Shih, John Tomaszewski and Anant Madabhushi. Automatic detection of invasive ductal carcinoma in whole slide images with Convolutional Neural Networks. Digital Pathology Conference. SPIE Medical Imaging 2014, 15 - 20 February 2014. Town & Country Resort and Convention Center, San Diego, California, USA. (Accepted) 72 Quantitative Histomorphometry to Identify Aggressive vs. Indolent p16+ Oropharyngeal Squamous Cell Carcinoma Image analysis of the tissue microarrays. H&E images of tumors from patients who either did not develop recurrent disease (A) or who went on to develop recurrent disease (D) (each 10X magnification). Nuclei were identified by computerized image analysis (B and E) and cell cluster graphs (C and F), shown by blue nodes for the nuclei with interconnecting red lines, generated. Insets of each panel show magnified areas of analysis. Preliminary results show Kaplan-Meier curves for 160 p16+ OSCC for (a) Tstage (T), (b) nodal status, and (c) QuHbIC. QuHbIC clearly outperforms T- and N-stage. Lewis, J, Ali, S, Luo, J, Thorstad, W, Madabhushi, A. A Quantitative Histomorphometric Classifier (QuHbIC) Identifies Aggressive Versus Indolent p16-Positive Oropharyngeal Squamous Cell Carcinoma. American Journal of Surgical Pathology: January 2014 - Volume 38 - Issue 1 - p 128–137 73 Cascaded Combination of CNN and Hand-crafted Features for Mitosis Detection – DWLBP CNN Convolutional neural networks Handcrafted features – Haralick – HOG – LBP – Run-length – Color histogram – Morphological false positives Cascaded ensemble 01/08/13 true positives false negatives Wang H, et al. “Cascaded ensemble of CNN and hand-crafted features for mitosis detection”, SPIE Medical imaging, 2014. 74 Spatio-temporal Texture (SpTeT) for Distinguishing Vulnerable from StableAtherosclerotic Plaque on Dynamic Contrast Enhanced MRI in a Rabbit Model Module 1: Data preprocessing Step 1: 3D registration Lumen Vessel outer wall T1wBB DCE-MRI DCE-MRI T1wBB Before Registration Step 2: Segmentation After Registration T1wBB T1wBB Vessel wall 3D visualization Two examples of 3D plots for (a): first-order textural feature (mean); (b) Haralick (intensity entropy); The figures showed that the SpTeT features are effective imaging based descriptors for differentiating vulnerable and stable plaques. DCE-MRI DCE-MRI 1 Module 2: Feature extraction Vessel wall Polynomial fitting Module 3: Classification Feature selection Random forests classifier 0 (a) Texture images 3D plotting (b) (c) (d) 1 ROC curve 0 Main components: Module 1 shows the preprocessing procedure to register and segment DCE-MRI and T1wBB. Module 2 extracts the SpTeT features, including Gabor, Kirsch, Sobel, Haralick, and first-order textural. In Module 3, quantitative evaluation of the features in distinguishing vulnerable from stable plaque on MRI was done via a random forests classifier. (e) (f) (g) (h) Examples of the contrast enhancement patterns associated with stable plaque (1st row) and vulnerable plaque (2nd row). (a), (e) and (b), (f): pre-contrast and postcontrast images at peak enhancement. (c) and (g) are the vessel and lumen segmentation. (d) and (h) are texture images using first-order texture feature. T.Wan, A. Phinikaridou, J.A.Hamilton, N.Hua, T. Pham, J.Danagoulian, R. Kleiman, A.J.Buckler, and A.Madabhushi, Spatio-temporal texture (SpTeT) for distinguishing vulnerable from stable atherosclerotic plaque on dynamic contrast enhancement (DCE) MRI in a rabbit model, submitted to Medical Physics, 2013. 75 Imaging Biomarkers for Stratifying ER+ Breast Cancer Risk We developed an image-based risk score (Ibris) using imaging biomarkers on biopsy samples to predict ER+ BCa risk (as determined by ODX). Nuclear Boundaries Median Value Sobel Filtered Image Morphological features from nuclear boundaries quantify pleomorphism (21 features) Texture features from inside nuclei quantify mitotic index (2,322 features) Supervised Classifier Performance Low Risk Patient (RS < 18) Metric Average (σ) AUC 0.87 (0.018) PPV 0.81 (0.039) NPV 0.88 (0.017) High Risk Patient (RS > 30) Madabhushi, A, Doyle, S, Basavanhally, A, Gilmore, H, Feldman, M, Mies, C, Tomaszewski, J, Harris, L, Ganesan, S, “Computer extracted image measurements of nuclear shape and texture from H&E images appear to stratify low and high risk ER+ breast cancers assessed via Oncotype DX“, 2013 San Antonio Breast Cancer Symposium, December 10- 14, 2013 in San Antonio, Texas 76 SOFTWARE 77 ProstaCAD Ver. 1 ProstaCAD is a Tumor Detection Software for T2 Images developed at CCIPD. Gives a simple means of visualization, manipulation and processing for 3D volume images. Extracts pre-specified set of Texture features from the given volume images. Classifies tumorous regions using discriminant analysis (LDA and QDA) 78 ProstaCAD Ver. 2 Supports Extended set of Texture Features with numerous customization settings. A lot of preprocessing and features have been added Supports ADC and DCE MRI Training module has been extended Validation module implementation is under progress. 79 CaPView Ver. 1 User-Friendly Graphical User interface for Prostate Cancer Probability Visualization Supports T2/ADC MRI. Supports 3D Visualization of the prostate surface as well as cancer probability heatmap in a fully customizable, yet simple, easy to understand environment. 80 INTERESTED IN JOINING CCIPD? We are always looking for enthusiastic and motivated graduate, undergraduate students and research scientists. If you think you would be a good fit for CCIPD, send over your complete CV and 3 representative publications to “anant.madabhushi” @ “case.edu” Follow us on Twitter: @CCIPD_Case 81