Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods
<italic>Goal:</italic> To predict physician fixations specifically on ophthalmology optical coherence tomography (OCT) reports from eye tracking data using CNN based saliency prediction methods in order to aid in the education of ophthalmologists and ophthalmologists-in-training. <ita...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2024-01-01
|
Series: | IEEE Open Journal of Engineering in Medicine and Biology |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10440538/ |
_version_ | 1797243373581500416 |
---|---|
author | Mingyang Zang Pooja Mukund Britney Forsyth Andrew F. Laine Kaveri A. Thakoor |
author_facet | Mingyang Zang Pooja Mukund Britney Forsyth Andrew F. Laine Kaveri A. Thakoor |
author_sort | Mingyang Zang |
collection | DOAJ |
description | <italic>Goal:</italic> To predict physician fixations specifically on ophthalmology optical coherence tomography (OCT) reports from eye tracking data using CNN based saliency prediction methods in order to aid in the education of ophthalmologists and ophthalmologists-in-training. <italic>Methods:</italic> Fifteen ophthalmologists were recruited to each examine 20 randomly selected OCT reports and evaluate the likelihood of glaucoma for each report on a scale of 0-100. Eye movements were collected using a Pupil Labs Core eye-tracker. Fixation heat maps were generated using fixation data. <italic>Results:</italic> A model trained with traditional saliency mapping resulted in a correlation coefficient (CC) value of 0.208, a Normalized Scanpath Saliency (NSS) value of 0.8172, a Kullback–Leibler (KLD) value of 2.573, and a Structural Similarity Index (SSIM) of 0.169. <italic>Conclusions</italic>: The TranSalNet model was able to predict fixations within certain regions of the OCT report with reasonable accuracy, but more data is needed to improve model accuracy. Future steps include increasing data collection, improving quality of data, and modifying the model architecture. |
first_indexed | 2024-04-24T18:54:05Z |
format | Article |
id | doaj.art-8c7851bcf8aa46cdb61f70ceef1897ba |
institution | Directory Open Access Journal |
issn | 2644-1276 |
language | English |
last_indexed | 2024-04-24T18:54:05Z |
publishDate | 2024-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Open Journal of Engineering in Medicine and Biology |
spelling | doaj.art-8c7851bcf8aa46cdb61f70ceef1897ba2024-03-26T17:46:30ZengIEEEIEEE Open Journal of Engineering in Medicine and Biology2644-12762024-01-01519119710.1109/OJEMB.2024.336749210440538Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction MethodsMingyang Zang0https://orcid.org/0009-0009-0830-6112Pooja Mukund1https://orcid.org/0009-0007-6396-114XBritney Forsyth2https://orcid.org/0009-0001-7214-7293Andrew F. Laine3https://orcid.org/0000-0003-3797-0628Kaveri A. Thakoor4https://orcid.org/0000-0001-5589-8151Columbia University, New York, NY, USAColumbia University, New York, NY, USAColumbia University, New York, NY, USAColumbia University, New York, NY, USAColumbia University, New York, NY, USA<italic>Goal:</italic> To predict physician fixations specifically on ophthalmology optical coherence tomography (OCT) reports from eye tracking data using CNN based saliency prediction methods in order to aid in the education of ophthalmologists and ophthalmologists-in-training. <italic>Methods:</italic> Fifteen ophthalmologists were recruited to each examine 20 randomly selected OCT reports and evaluate the likelihood of glaucoma for each report on a scale of 0-100. Eye movements were collected using a Pupil Labs Core eye-tracker. Fixation heat maps were generated using fixation data. <italic>Results:</italic> A model trained with traditional saliency mapping resulted in a correlation coefficient (CC) value of 0.208, a Normalized Scanpath Saliency (NSS) value of 0.8172, a Kullback–Leibler (KLD) value of 2.573, and a Structural Similarity Index (SSIM) of 0.169. <italic>Conclusions</italic>: The TranSalNet model was able to predict fixations within certain regions of the OCT report with reasonable accuracy, but more data is needed to improve model accuracy. Future steps include increasing data collection, improving quality of data, and modifying the model architecture.https://ieeexplore.ieee.org/document/10440538/Deep learningoptical coherence tomographysaliency prediction |
spellingShingle | Mingyang Zang Pooja Mukund Britney Forsyth Andrew F. Laine Kaveri A. Thakoor Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods IEEE Open Journal of Engineering in Medicine and Biology Deep learning optical coherence tomography saliency prediction |
title | Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods |
title_full | Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods |
title_fullStr | Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods |
title_full_unstemmed | Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods |
title_short | Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods |
title_sort | predicting clinician fixations on glaucoma oct reports via cnn based saliency prediction methods |
topic | Deep learning optical coherence tomography saliency prediction |
url | https://ieeexplore.ieee.org/document/10440538/ |
work_keys_str_mv | AT mingyangzang predictingclinicianfixationsonglaucomaoctreportsviacnnbasedsaliencypredictionmethods AT poojamukund predictingclinicianfixationsonglaucomaoctreportsviacnnbasedsaliencypredictionmethods AT britneyforsyth predictingclinicianfixationsonglaucomaoctreportsviacnnbasedsaliencypredictionmethods AT andrewflaine predictingclinicianfixationsonglaucomaoctreportsviacnnbasedsaliencypredictionmethods AT kaveriathakoor predictingclinicianfixationsonglaucomaoctreportsviacnnbasedsaliencypredictionmethods |