Hydranet: Data augmentation for regression neural networks

TY - GEN

T1 - Hydranet

T2 - 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2019

AU - Dubost, Florian

AU - Bortsova, Gerda

AU - Adams, Hieab

AU - Ikram, M. Arfan

AU - Niessen, Wiro

AU - Vernooij, Meike

AU - de Bruijne, Marleen

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Deep learning techniques are often criticized to heavily depend on a large quantity of labeled data. This problem is even more challenging in medical image analysis where the annotator expertise is often scarce. We propose a novel data-augmentation method to regularize neural network regressors that learn from a single global label per image. The principle of the method is to create new samples by recombining existing ones. We demonstrate the performance of our algorithm on two tasks: estimation of the number of enlarged perivascular spaces in the basal ganglia, and estimation of white matter hyperintensities volume. We show that the proposed method improves the performance over more basic data augmentation. The proposed method reached an intraclass correlation coefficient between ground truth and network predictions of 0.73 on the first task and 0.84 on the second task, only using between 25 and 30 scans with a single global label per scan for training. With the same number of training scans, more conventional data augmentation methods could only reach intraclass correlation coefficients of 0.68 on the first task, and 0.79 on the second task.

AB - Deep learning techniques are often criticized to heavily depend on a large quantity of labeled data. This problem is even more challenging in medical image analysis where the annotator expertise is often scarce. We propose a novel data-augmentation method to regularize neural network regressors that learn from a single global label per image. The principle of the method is to create new samples by recombining existing ones. We demonstrate the performance of our algorithm on two tasks: estimation of the number of enlarged perivascular spaces in the basal ganglia, and estimation of white matter hyperintensities volume. We show that the proposed method improves the performance over more basic data augmentation. The proposed method reached an intraclass correlation coefficient between ground truth and network predictions of 0.73 on the first task and 0.84 on the second task, only using between 25 and 30 scans with a single global label per scan for training. With the same number of training scans, more conventional data augmentation methods could only reach intraclass correlation coefficients of 0.68 on the first task, and 0.79 on the second task.

UR - http://www.scopus.com/inward/record.url?scp=85075675428&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-32251-9_48

DO - 10.1007/978-3-030-32251-9_48

M3 - Article in proceedings

AN - SCOPUS:85075675428

SN - 9783030322502

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 438

EP - 446

BT - Medical Image Computing and Computer Assisted Intervention – MICCAI 2019 - 22nd International Conference, Proceedings

A2 - Shen, Dinggang

A2 - Yap, Pew-Thian

A2 - Liu, Tianming

A2 - Peters, Terry M.

A2 - Khan, Ali

A2 - Staib, Lawrence H.

A2 - Essert, Caroline

A2 - Zhou, Sean

PB - Springer VS

Y2 - 13 October 2019 through 17 October 2019

ER -