Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy
Optical coherence tomography angiography (OCTA) can deliver enhanced diagnosis for diabetic retinopathy (DR). This study evaluated a deep learning (DL) algorithm for automatic DR severity assessment using high-resolution and ultra-widefield (UWF) OCTA. Diabetic patients were examined with <inline...
Main Authors: | , , , , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-08-01
|
Series: | Diagnostics |
Subjects: | |
Online Access: | https://www.mdpi.com/2075-4418/13/17/2770 |
_version_ | 1797582657154973696 |
---|---|
author | Yihao Li Mostafa El Habib Daho Pierre-Henri Conze Rachid Zeghlache Hugo Le Boité Sophie Bonnin Deborah Cosette Stephanie Magazzeni Bruno Lay Alexandre Le Guilcher Ramin Tadayoni Béatrice Cochener Mathieu Lamard Gwenolé Quellec |
author_facet | Yihao Li Mostafa El Habib Daho Pierre-Henri Conze Rachid Zeghlache Hugo Le Boité Sophie Bonnin Deborah Cosette Stephanie Magazzeni Bruno Lay Alexandre Le Guilcher Ramin Tadayoni Béatrice Cochener Mathieu Lamard Gwenolé Quellec |
author_sort | Yihao Li |
collection | DOAJ |
description | Optical coherence tomography angiography (OCTA) can deliver enhanced diagnosis for diabetic retinopathy (DR). This study evaluated a deep learning (DL) algorithm for automatic DR severity assessment using high-resolution and ultra-widefield (UWF) OCTA. Diabetic patients were examined with <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>6</mn><mo>×</mo><mn>6</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> high-resolution OCTA and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>15</mn><mo>×</mo><mn>15</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> UWF-OCTA using PLEX®Elite 9000. A novel DL algorithm was trained for automatic DR severity inference using both OCTA acquisitions. The algorithm employed a unique hybrid fusion framework, integrating structural and flow information from both acquisitions. It was trained on data from 875 eyes of 444 patients. Tested on 53 patients (97 eyes), the algorithm achieved a good area under the receiver operating characteristic curve (AUC) for detecting DR (0.8868), moderate non-proliferative DR (0.8276), severe non-proliferative DR (0.8376), and proliferative/treated DR (0.9070). These results significantly outperformed detection with the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>6</mn><mo>×</mo><mn>6</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> (AUC = 0.8462, 0.7793, 0.7889, and 0.8104, respectively) or <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>15</mn><mo>×</mo><mn>15</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> (AUC = 0.8251, 0.7745, 0.7967, and 0.8786, respectively) acquisitions alone. Thus, combining high-resolution and UWF-OCTA acquisitions holds the potential for improved early and late-stage DR detection, offering a foundation for enhancing DR management and a clear path for future works involving expanded datasets and integrating additional imaging modalities. |
first_indexed | 2024-03-10T23:25:35Z |
format | Article |
id | doaj.art-425ebe766038431b9d142efc946ffaaf |
institution | Directory Open Access Journal |
issn | 2075-4418 |
language | English |
last_indexed | 2024-03-10T23:25:35Z |
publishDate | 2023-08-01 |
publisher | MDPI AG |
record_format | Article |
series | Diagnostics |
spelling | doaj.art-425ebe766038431b9d142efc946ffaaf2023-11-19T07:59:19ZengMDPI AGDiagnostics2075-44182023-08-011317277010.3390/diagnostics13172770Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic RetinopathyYihao Li0Mostafa El Habib Daho1Pierre-Henri Conze2Rachid Zeghlache3Hugo Le Boité4Sophie Bonnin5Deborah Cosette6Stephanie Magazzeni7Bruno Lay8Alexandre Le Guilcher9Ramin Tadayoni10Béatrice Cochener11Mathieu Lamard12Gwenolé Quellec13Inserm, UMR 1101 LaTIM, F-29200 Brest, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceSorbonne University, F-75006 Paris, FranceService d’Ophtalmologie, Hôpital Lariboisière, AP-HP, F-75475 Paris, FranceCarl Zeiss Meditec Inc., Dublin, CA 94568, USACarl Zeiss Meditec Inc., Dublin, CA 94568, USAADCIS, F-14280 Saint-Contest, FranceEvolucare Technologies, F-78230 Le Pecq, FranceService d’Ophtalmologie, Hôpital Lariboisière, AP-HP, F-75475 Paris, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceInserm, UMR 1101 LaTIM, F-29200 Brest, FranceOptical coherence tomography angiography (OCTA) can deliver enhanced diagnosis for diabetic retinopathy (DR). This study evaluated a deep learning (DL) algorithm for automatic DR severity assessment using high-resolution and ultra-widefield (UWF) OCTA. Diabetic patients were examined with <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>6</mn><mo>×</mo><mn>6</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> high-resolution OCTA and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>15</mn><mo>×</mo><mn>15</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> UWF-OCTA using PLEX®Elite 9000. A novel DL algorithm was trained for automatic DR severity inference using both OCTA acquisitions. The algorithm employed a unique hybrid fusion framework, integrating structural and flow information from both acquisitions. It was trained on data from 875 eyes of 444 patients. Tested on 53 patients (97 eyes), the algorithm achieved a good area under the receiver operating characteristic curve (AUC) for detecting DR (0.8868), moderate non-proliferative DR (0.8276), severe non-proliferative DR (0.8376), and proliferative/treated DR (0.9070). These results significantly outperformed detection with the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>6</mn><mo>×</mo><mn>6</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> (AUC = 0.8462, 0.7793, 0.7889, and 0.8104, respectively) or <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>15</mn><mo>×</mo><mn>15</mn></mrow></semantics></math></inline-formula> mm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>2</mn></msup></semantics></math></inline-formula> (AUC = 0.8251, 0.7745, 0.7967, and 0.8786, respectively) acquisitions alone. Thus, combining high-resolution and UWF-OCTA acquisitions holds the potential for improved early and late-stage DR detection, offering a foundation for enhancing DR management and a clear path for future works involving expanded datasets and integrating additional imaging modalities.https://www.mdpi.com/2075-4418/13/17/2770diabetic retinopathy classificationmultimodal information fusiondeep learningcomputer-aided diagnosis |
spellingShingle | Yihao Li Mostafa El Habib Daho Pierre-Henri Conze Rachid Zeghlache Hugo Le Boité Sophie Bonnin Deborah Cosette Stephanie Magazzeni Bruno Lay Alexandre Le Guilcher Ramin Tadayoni Béatrice Cochener Mathieu Lamard Gwenolé Quellec Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy Diagnostics diabetic retinopathy classification multimodal information fusion deep learning computer-aided diagnosis |
title | Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy |
title_full | Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy |
title_fullStr | Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy |
title_full_unstemmed | Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy |
title_short | Hybrid Fusion of High-Resolution and Ultra-Widefield OCTA Acquisitions for the Automatic Diagnosis of Diabetic Retinopathy |
title_sort | hybrid fusion of high resolution and ultra widefield octa acquisitions for the automatic diagnosis of diabetic retinopathy |
topic | diabetic retinopathy classification multimodal information fusion deep learning computer-aided diagnosis |
url | https://www.mdpi.com/2075-4418/13/17/2770 |
work_keys_str_mv | AT yihaoli hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT mostafaelhabibdaho hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT pierrehenriconze hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT rachidzeghlache hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT hugoleboite hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT sophiebonnin hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT deborahcosette hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT stephaniemagazzeni hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT brunolay hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT alexandreleguilcher hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT ramintadayoni hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT beatricecochener hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT mathieulamard hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy AT gwenolequellec hybridfusionofhighresolutionandultrawidefieldoctaacquisitionsfortheautomaticdiagnosisofdiabeticretinopathy |