News

Tue, 2017-06-13 10:29
ML-CDS 2017: Multimodal Learning for Clinical Decision Support
Tuesday, June 13, 2017 - 10:29

NEW deadline for paper submission: June 26, 2017

In conjunction with MICCAI 2017

September 10, 2017,Quebec City, Canada

http://www.mcbr-cds.org/

Now accepting submissions



Papers will be published in a volume of Springer LNCS series (MICCAI 2017 Workshops Volume) 

 

ML-CDS 2017 builds on the success of the last four events in this series. We are looking for high-quality submissions that address innovative research and development in the learning methods using multimodal medical data. Applications in clinical decision support and treatment planning are of highest interest. Experts in quantitative imaging, text analytics, and decision support systems will present to an audience of scientists and clinicians. Advances in the development and use of deep learning methods with medical imaging and text data are expected to be among major topics of submission and discussion at the event.

Tue, 2016-11-15 11:43
Ania Gawlik's article featured in the Best of the 2016 AUA
Tuesday, November 15, 2016 - 11:43

Ania Gawlik's article was featured in a paper describing the best of AUA. The quoted text is as follows.

"In the 1970s there was great interest in prostate cancer cytology (based on needle aspiration) for prostate cancer diagnoses. It was supplanted after the pioneering work of Gleason demonstrated that the architecture of prostate cancer histology provided very significant prognostic information. History was reassessed in the report from Gawlik and coworkers,93 who utilized computer image analysis of Feulgen and hematoxylin-eosin (H&E) nuclear features to predict BCR in men following RP; 69 patients (20 BCR and 49 nonrecurrences) were assessed with mean BCR-free survival time of 6.6 years and follow-up to 14 years. A total of 242 quantitative histomorphometric (QH) features describing nuclear shape, architecture, and disorder were calculated from the H&E and Feulgen-stained tissue microarray core images in each patient. The top 10 ranked features for each stain type were selected.

Gleason score did not discriminate between those who did or did not have BCR predictions using QH features extracted from Feulgen and H&E images and revealed statistically different outcomes. Combining Gleason score, H&E, and Feulgen together showed the highest classification accuracy (0.75; P < .001). Although this is a very small study, if validated, this may provide a fruitful arena for development—perhaps there is something new (again) under the sun."

 

The full paper can be found at 

Rev Urol. 2016; 18(3): 159–173.
Highlights From the 2016 American Urological Association Annual Meeting, May 6–10, 2016, San Diego, CA

Citation of CCIPD's paper is 

Gawlik A, Lee G, Whitney J, et al. Computer extracted nuclear features from Feulgen and H&E images predict biochemical recurrence in prostate cancer patients following radical prostatectomy [AUA abstract MP02-17] J Urol. 2016;195(suppl 4):e16–e17.

Tue, 2016-11-08 11:11
Five abstracts accepted for USCAP
Tuesday, November 8, 2016 - 11:11

Five abstracts from CCIPD have been accepted for USCAP's 106th Annual Meeting. The meeting will be held in San Antonio, TX from March 4th to 10th, 2017.

The following are the abstracts that have been selected

  • “Deep Learning Automated Segmentation of Tumor Infiltrating Lymphocytes in Breast Cancer Specimens”

  • “Deep Learning Classifier to Predict Cardiac Failure from Whole-Slide H&E Images”

  • “Computerized Density Estimation of Tumor-Infiltrating Lymphocyte in H&E TMAs Predicts Recurrence in Early Stage Non-Small Cell Lung Cancer”

  • “Computer Extracted Features of Nuclear Architecture in H&E Sectionsare Predictive of Disease Specific Survival Inoral Cavity Squamous Cell Carcinoma Patients”

  • “Identifying the Histomorphometric Basis of MRI Radiomic Features in Distinguishing Gleason Grades of Prostate Cancer”

 

Fri, 2016-11-04 10:11
New patent issued to CCIPD/BrIC labs
Friday, November 4, 2016 - 10:11

 

US Patent US 9,483,822 "CO-OCCURRENCE OF LOCAL ANISOTROPIC GRADIENT ORIENTATIONS" has been issued with Co-inventors Anant Madabhushi, Prateek Prasanna and Pallavi Tiwari.
 
Abstract of the invention is as follows:-
 
Methods, apparatus, and other embodiments associated with distinguishing disease phenotypes using co-occurrence of local anisotropic gradient orientations (CoLIAGe) are described. One example apparatus includes a set of logics that acquires a radiologic image (e.g., MRI image) of a region of tissue demonstrating disease pathology (e.g., cancer), computes a gradient orientation for a pixel in the MRI image, computes a significant orientation for the pixel based on the gradient orientation, constructs a feature vector that captures a discretized entropy distribution for the image based on the significant orientation, and classifies the phenotype of the disease pathology based on the feature vector. Embodiments of example apparatus may generate and display a heatmap of entropy values for the image.

US Patent US 9,483,822 "CO-OCCURRENCE OF LOCAL ANISOTROPIC GRADIENT ORIENTATIONS" has been issued with Co-inventors Anant Madabhushi, Prateek Prasanna and Pallavi Tiwari.
 
Abstract of the invention is as follows:-
 
Methods, apparatus, and other embodiments associated with distinguishing disease phenotypes using co-occurrence of local anisotropic gradient orientations (CoLIAGe) are described. One example apparatus includes a set of logics that acquires a radiologic image (e.g., MRI image) of a region of tissue demonstrating disease pathology (e.g., cancer), computes a gradient orientation for a pixel in the MRI image, computes a significant orientation for the pixel based on the gradient orientation, constructs a feature vector that captures a discretized entropy distribution for the image based on the significant orientation, and classifies the phenotype of the disease pathology based on the feature vector. Embodiments of example apparatus may generate and display a heatmap of entropy values for the image.