Improved interpretability for computer-aided severity assessment of retinopathy of prematurity

Mara Graziani, James M. Brown, Vincent Andrearczyk, Veysi Yildiz, John Campbell, Deniz Erdogmus, Stratis Ioannidis, Michael Chiang, Jayashree Kalpathy-Cramer, Henning Müller

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Computer-aided diagnosis tools for Retinopathy of Prematurity (ROP) base their decisions on handcrafted retinal features that highly correlate with expert diagnoses, such as arterial and venous curvature, tortuosity and dilation. Deep learning leads to performance comparable to those of expert physicians, albeit not ensuring that the same clinical factors are learned in the deep representations. In this paper, we investigate the relationship between the handcrafted and the deep learning features in the context of ROP diagnosis. Average statistics on the handcrafted features for each input image were expressed as retinal concept measures. Three disease severity grades, i.e. normal, pre-plus and plus, were classified by a deep convolutional neural network. Regression Concept Vectors (RCV) were computed in the network feature space for each retinal concept measure. Relevant concept measures were identified by bidirectional relevance scores for the normal and plus classes. Results show that the curvature, diameter and tortuosity of the segmented vessels are indeed relevant to the classification. Among the potential applications of this method, the analysis of borderline cases between the classes and of network faults, in particular, can be used to improve the performance.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2019
Subtitle of host publicationComputer-Aided Diagnosis
EditorsKensaku Mori, Horst K. Hahn
PublisherSPIE
ISBN (Electronic)9781510625471
DOIs
StatePublished - Jan 1 2019
EventMedical Imaging 2019: Computer-Aided Diagnosis - San Diego, United States
Duration: Feb 17 2019Feb 20 2019

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume10950
ISSN (Print)1605-7422

Conference

ConferenceMedical Imaging 2019: Computer-Aided Diagnosis
CountryUnited States
CitySan Diego
Period2/17/192/20/19

Fingerprint

Retinopathy of Prematurity
Computer aided diagnosis
learning
curvature
Learning
physicians
Statistics
Neural networks
vessels
regression analysis
Dilatation
grade
statistics
Physicians
Deep learning

Keywords

  • Deep learning
  • Interpretability
  • Machine learning
  • Plus disease
  • Retinopathy

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Atomic and Molecular Physics, and Optics
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Graziani, M., Brown, J. M., Andrearczyk, V., Yildiz, V., Campbell, J., Erdogmus, D., ... Müller, H. (2019). Improved interpretability for computer-aided severity assessment of retinopathy of prematurity. In K. Mori, & H. K. Hahn (Eds.), Medical Imaging 2019: Computer-Aided Diagnosis [109501R] (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10950). SPIE. https://doi.org/10.1117/12.2512584

Improved interpretability for computer-aided severity assessment of retinopathy of prematurity. / Graziani, Mara; Brown, James M.; Andrearczyk, Vincent; Yildiz, Veysi; Campbell, John; Erdogmus, Deniz; Ioannidis, Stratis; Chiang, Michael; Kalpathy-Cramer, Jayashree; Müller, Henning.

Medical Imaging 2019: Computer-Aided Diagnosis. ed. / Kensaku Mori; Horst K. Hahn. SPIE, 2019. 109501R (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10950).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Graziani, M, Brown, JM, Andrearczyk, V, Yildiz, V, Campbell, J, Erdogmus, D, Ioannidis, S, Chiang, M, Kalpathy-Cramer, J & Müller, H 2019, Improved interpretability for computer-aided severity assessment of retinopathy of prematurity. in K Mori & HK Hahn (eds), Medical Imaging 2019: Computer-Aided Diagnosis., 109501R, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 10950, SPIE, Medical Imaging 2019: Computer-Aided Diagnosis, San Diego, United States, 2/17/19. https://doi.org/10.1117/12.2512584
Graziani M, Brown JM, Andrearczyk V, Yildiz V, Campbell J, Erdogmus D et al. Improved interpretability for computer-aided severity assessment of retinopathy of prematurity. In Mori K, Hahn HK, editors, Medical Imaging 2019: Computer-Aided Diagnosis. SPIE. 2019. 109501R. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE). https://doi.org/10.1117/12.2512584
Graziani, Mara ; Brown, James M. ; Andrearczyk, Vincent ; Yildiz, Veysi ; Campbell, John ; Erdogmus, Deniz ; Ioannidis, Stratis ; Chiang, Michael ; Kalpathy-Cramer, Jayashree ; Müller, Henning. / Improved interpretability for computer-aided severity assessment of retinopathy of prematurity. Medical Imaging 2019: Computer-Aided Diagnosis. editor / Kensaku Mori ; Horst K. Hahn. SPIE, 2019. (Progress in Biomedical Optics and Imaging - Proceedings of SPIE).
@inproceedings{ef3ba28a3a744d25bd29908fadce8aa9,
title = "Improved interpretability for computer-aided severity assessment of retinopathy of prematurity",
abstract = "Computer-aided diagnosis tools for Retinopathy of Prematurity (ROP) base their decisions on handcrafted retinal features that highly correlate with expert diagnoses, such as arterial and venous curvature, tortuosity and dilation. Deep learning leads to performance comparable to those of expert physicians, albeit not ensuring that the same clinical factors are learned in the deep representations. In this paper, we investigate the relationship between the handcrafted and the deep learning features in the context of ROP diagnosis. Average statistics on the handcrafted features for each input image were expressed as retinal concept measures. Three disease severity grades, i.e. normal, pre-plus and plus, were classified by a deep convolutional neural network. Regression Concept Vectors (RCV) were computed in the network feature space for each retinal concept measure. Relevant concept measures were identified by bidirectional relevance scores for the normal and plus classes. Results show that the curvature, diameter and tortuosity of the segmented vessels are indeed relevant to the classification. Among the potential applications of this method, the analysis of borderline cases between the classes and of network faults, in particular, can be used to improve the performance.",
keywords = "Deep learning, Interpretability, Machine learning, Plus disease, Retinopathy",
author = "Mara Graziani and Brown, {James M.} and Vincent Andrearczyk and Veysi Yildiz and John Campbell and Deniz Erdogmus and Stratis Ioannidis and Michael Chiang and Jayashree Kalpathy-Cramer and Henning M{\"u}ller",
year = "2019",
month = "1",
day = "1",
doi = "10.1117/12.2512584",
language = "English (US)",
series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
publisher = "SPIE",
editor = "Kensaku Mori and Hahn, {Horst K.}",
booktitle = "Medical Imaging 2019",

}

TY - GEN

T1 - Improved interpretability for computer-aided severity assessment of retinopathy of prematurity

AU - Graziani, Mara

AU - Brown, James M.

AU - Andrearczyk, Vincent

AU - Yildiz, Veysi

AU - Campbell, John

AU - Erdogmus, Deniz

AU - Ioannidis, Stratis

AU - Chiang, Michael

AU - Kalpathy-Cramer, Jayashree

AU - Müller, Henning

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Computer-aided diagnosis tools for Retinopathy of Prematurity (ROP) base their decisions on handcrafted retinal features that highly correlate with expert diagnoses, such as arterial and venous curvature, tortuosity and dilation. Deep learning leads to performance comparable to those of expert physicians, albeit not ensuring that the same clinical factors are learned in the deep representations. In this paper, we investigate the relationship between the handcrafted and the deep learning features in the context of ROP diagnosis. Average statistics on the handcrafted features for each input image were expressed as retinal concept measures. Three disease severity grades, i.e. normal, pre-plus and plus, were classified by a deep convolutional neural network. Regression Concept Vectors (RCV) were computed in the network feature space for each retinal concept measure. Relevant concept measures were identified by bidirectional relevance scores for the normal and plus classes. Results show that the curvature, diameter and tortuosity of the segmented vessels are indeed relevant to the classification. Among the potential applications of this method, the analysis of borderline cases between the classes and of network faults, in particular, can be used to improve the performance.

AB - Computer-aided diagnosis tools for Retinopathy of Prematurity (ROP) base their decisions on handcrafted retinal features that highly correlate with expert diagnoses, such as arterial and venous curvature, tortuosity and dilation. Deep learning leads to performance comparable to those of expert physicians, albeit not ensuring that the same clinical factors are learned in the deep representations. In this paper, we investigate the relationship between the handcrafted and the deep learning features in the context of ROP diagnosis. Average statistics on the handcrafted features for each input image were expressed as retinal concept measures. Three disease severity grades, i.e. normal, pre-plus and plus, were classified by a deep convolutional neural network. Regression Concept Vectors (RCV) were computed in the network feature space for each retinal concept measure. Relevant concept measures were identified by bidirectional relevance scores for the normal and plus classes. Results show that the curvature, diameter and tortuosity of the segmented vessels are indeed relevant to the classification. Among the potential applications of this method, the analysis of borderline cases between the classes and of network faults, in particular, can be used to improve the performance.

KW - Deep learning

KW - Interpretability

KW - Machine learning

KW - Plus disease

KW - Retinopathy

UR - http://www.scopus.com/inward/record.url?scp=85068128996&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068128996&partnerID=8YFLogxK

U2 - 10.1117/12.2512584

DO - 10.1117/12.2512584

M3 - Conference contribution

AN - SCOPUS:85068128996

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2019

A2 - Mori, Kensaku

A2 - Hahn, Horst K.

PB - SPIE

ER -