Click here to


Are you sure ?

Yes, do it No, cancel

Characterization of Local Spatial Information Utilized in Training of Deep Convolutional Neural Networks (DCNN) for Automatic Segmentation of the Prostate On CT Images

C Liu*, S Gardner , N Wen , E Mohamed , B Movsas , I Chetty , Henry Ford Health System, Detroit, MI


(Wednesday, 7/17/2019) 10:00 AM - 10:30 AM

Room: Exhibit Hall | Forum 2

Purpose: The importance of spatial image information utilized by a DCNN in automatic segmentation is central toward understanding how to train the DCNN for optimal performance. Here we evaluate the behavior of a DCNN for prostate segmentation on CT images using the Local Interpretable Model-agnostic Explanations (LIME), technique [1], which enables generation of an interpretable model by characterizing local sub-groups/clusters contained within the image.

Methods: Planning-CT (pCT) datasets for 1104 prostate cancer patients were retrospectively selected. Nine-hundred-sixty-four datasets were used for training/validation and 140 were used for testing. All images were resampled to spatial resolution of 1x1x1.5mm, and a DCNN was trained. The top performing DCNN was chosen based on validation results and used to auto-segment the prostate on all testing images. Results were compared between DNN and physician-generated contours using Dice coefficient(DSC). The importance of each sub-region was evaluated using forward feature selection following the LIME method, and 100 (of 4K) sub-regions were used for final classification. Multiple experiments were carried out to select sub-regions using different numbers of random samples. The optimal parameter was determined when the selected sub-regions converged. Each of the 140 testing datasets was characterized using the same parameters.

Results: Selected sub-regions converged at > 10K random samples. Altogether 1058 sub-regions were selected, out of which 775(73%) appeared more than once and 23(2%) appeared more than 70 times (50% of the 140 testing datasets). One sub-region was selected 91 times. Highest frequency sub-regions were observed to closest to the prostate gland, bladder and rectum.

Conclusion: The behavior of a DCNN based prostate segmentation algorithm was characterized/explained using a group of sub-regions on a per-testing sample basis. Characterization was consistent across datasets. Results showed that DCNN-based segmentation was associated primarily with image information in close vicinity of the prostate gland, bladder and rectum.

Funding Support, Disclosures, and Conflict of Interest: This work is supported in part by Varian Medical Systems, Palo Alto, CA.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email