Skip to main content

Medical-image Analysis and Statistical Interpretation Lab

Body Part Regression With Self-Supervision

Posted by on Thursday, January 14, 2021 in Body-Wise, Computed Tomography, Deep Learning, Machine Learning.

Y.Tang, R.Gao, S.Han, Y.Chen, D.Gao, V.Nath, C.Bermudez, M.R. Savona, R.G. Abramson, S.Bao,I.Lyu, Y.Huo and B.A. Landman,”Body Part Regression with Self-supervision”,IEEETransactions onMedicalImaging,2021

Full Text: 

https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9350603

Abstract

Body part regression is a promising new technique that enables content navigation through selfsupervised learning. Using this technique, the global quantitative spatial location for each axial view slice is obtained from computed tomography (CT). However, it is challenging to define a unified global coordinate system for body CT scans due to the large variabilities in image resolution, contrasts, sequences, and patient anatomy. Therefore, the widely used supervised learning approach cannot be easily deployed. To address these concerns, we propose an annotation-free method named blind-unsupervised-supervision network (BUSN). The contributions of the work are in four folds: (1) 1030 multi-center CT scans are used in developing BUSN without any manual annotation. (2) the proposed BUSN corrects the predictions from unsupervised learning and uses the corrected results as the new supervision; (3) to improve the consistency of predictions, we propose a novel neighbor message passing (NMP) scheme that is integrated with BUSN as a statistical learning based correction; and (4) we introduce a new pre-processing pipeline with inclusion of the BUSN, which is validated on 3D multi-organ segmentation. The proposed method is trained on 1,030 whole body CT scans (230,650 slices) from five datasets,as well as an independentexternal validation cohort with 100 scans. From the body part regression results, the proposed BUSN achieved significantly higher median R squared score ( = 0.9089) than the state-ofthe-art unsupervised method (= 0.7153). When introducing BUSN as a preprocessingstage in volumetric segmentation, the proposedpre-processingpipeline using BUSN approach increases the total mean Dice score of the 3D abdominal multi-organ segmentation from 0.7991 to 0.8145.

A representative subject was evaluated with URN (left) and BUSN (right). Green scatters are inliers of influence to the regression, yellow scatters are outliers of no influence to the distributed data. Darker blue line indicates the normal linear regression on scatters points, lighter blue line is the RANSAC regressor result according to inliers. Left panel presents the single URN regression with amounts of outliers result in failure of linearity nature in chest and pelvis regions. Right panel shows the testing result of BUSN method, the distributed scores follows good linearity in chest, abdomen and pelvis regions in CT scan. In summary, BUSN takes advantage of self-supervised network, which presents better continuity in regression result among neighbor slices and shows scatter plots without number of outliers.
A representative subject was evaluated with URN (left) and BUSN (right). Green scatters are inliers of influence to the regression, yellow scatters are outliers of no influence to the distributed data. Darker blue line indicates the normal linear regression on scatters points, lighter blue line is the RANSAC regressor result according to inliers. Left panel presents the single URN regression with amounts of outliers result in failure of linearity nature in chest and pelvis regions. Right panel shows the testing result of BUSN method, the distributed scores follows good linearity in chest, abdomen and pelvis regions in CT scan. In summary, BUSN takes advantage of self-supervised network, which presents better continuity in regression result among neighbor slices and shows scatter plots without number of outliers.

Tags: ,