Multimodal Deep Learning Model of Predicting Future Visual Field for Glaucoma Patients

被引：3

作者：

Pham, Quang T. M. ^{[1
]}

Han, Jong Chul ^{[2
,3
]}

Park, Do Young ^{[4
]}

Shin, Jitae ^{[1
]}

机构：

[1] Sungkyunkwan Univ, Dept Elect & Comp Engn, Suwon 16419, South Korea

[2] Sungkyunkwan Univ, Samsung Med Ctr, Dept Ophthalmol, Sch Med, Seoul 03181, South Korea

[3] Sungkyunkwan Univ, Dept Med Device Management & Res, SAIHST, Seoul 03181, South Korea

[4] Yeungnam Univ, Yeungnam Univ Hosp, Dept Ophthalmol, Coll Med, Daegu 42415, South Korea

来源：

IEEE ACCESS | 2023年 / 11卷

基金：

新加坡国家研究基金会;

关键词：

Predictive models; Noise measurement; Optical imaging; Deep learning; Analytical models; Optical fibers; Glaucoma; Visualization; glaucoma; OCT; visual field; QUALITY-OF-LIFE; OPTICAL COHERENCE TOMOGRAPHY; NERVE-FIBER LAYER; THICKNESS MAPS;

D O I：

10.1109/ACCESS.2023.3248065

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Glaucoma is one of the most common reasons for blindness worldwide, especially in elderly people. Glaucoma can be monitored using visual field (VF) tests. Therefore, predicting the future VF to monitor progression of glaucoma is important. In this paper, we proposed a deep learning model to predict future VF based on previous VF and optical coherence tomography (OCT) images (including thickness map, vertical tomogram, and horizontal tomogram). The image data were analyzed using a ResNet-50 model. Image features and previous VFs were combined, and a long short-term memory (LSTM) network was used to predict future VF. A weighted method was used to detect noisy data. The proposed method was improved when applying weighted loss. The mean absolute error (MAE) was 3.31 +/- 1.37, and the root mean square error (RMSE) was 4.58 +/- 1.77. The model showed high performance when combining VF data and OCT image data. Furthermore, the model was useful for detecting and re-weighting noisy data.

引用

页码：19049 / 19058

页数：10