Automated machine learning may fast detect visual field loss patterns in glaucoma

Published On 2022-08-01 14:30 GMT   |   Update On 2022-08-01 14:31 GMT

In a new study conducted by Siamak Yousefi and colleagues, it was found that an automated machine learning method can detect patterns of visual field (VF) loss and provide objective, reproducible terminology for describing early indicators of visual abnormalities and rapid progression in glaucoma patients. The findings of this study were published in Ophthalmology.This was a cross-sectional...

Login or Register to read the full article

In a new study conducted by Siamak Yousefi and colleagues, it was found that an automated machine learning method can detect patterns of visual field (VF) loss and provide objective, reproducible terminology for describing early indicators of visual abnormalities and rapid progression in glaucoma patients. The findings of this study were published in Ophthalmology.

This was a cross-sectional and longitudinal study that followed 2231 aberrant VFs from 205 eyes of 176 OHTS individuals for almost 16 years. An unsupervised deep archetypal analysis method and an OHTS certified VF reader were used to discover common patterns of VF loss. Machine-identified glaucoma damage patterns were compared to those previously described (expert-identified) in the OHTS in 2003. The longitudinal VFs of each eye were used to identify VF loss patterns that were highly related to rapid glaucoma progression. The key findings of this study were machine-expert correlation and the types of patterns of VF loss linked with rapid growth.

The key findings of this study were as follows:

1. The average VF mean deviation (MD) at glaucoma conversion was -2.7 dB (SD = 2.4 dB), while the mean MD of the eyes at the previous visit was -5.2 dB (SD = 5.5 dB).

2. Fifty of the 205 eyes exhibited an MD rate of -1 dB/year or more and were classified as quick progressors.

3. Eighteen machine-identified VF loss patterns were compared to expert-identified patterns, with 13 VF loss patterns being identical.

4. The most common expert-identified patterns were temporal wedge, partial arcuate, nasal step, and paracentral VF flaws, while the most common machine-identified patterns were temporal wedge, partial arcuate, nasal step, and paracentral VF defects.

5. After adjusting for gender, age, and starting MD, one of the machine-identified patterns of VF loss indicated future rapid VF advancement.

In conclusion, the creation of a machine-learning artificial intelligence system would really be extremely beneficial in clinical care by objectively identifying VF abnormalities related with quicker glaucoma progression.

Reference: 

Yousefi, S., Pasquale, L. R., Boland, M. V., & Johnson, C. A. (2022). Machine-identified Patterns of Visual Field Loss and An Association with Rapid Progression in the Ocular Hypertension Treatment Study. In Ophthalmology. Elsevier BV. https://doi.org/10.1016/j.ophtha.2022.07.001

Tags:    
Advertisement
Article Source : American Academy of Ophthalmology

Disclaimer: This site is primarily intended for healthcare professionals. Any content/information on this website does not replace the advice of medical and/or health professionals and should not be construed as medical/diagnostic advice/endorsement or prescription. Use of this site is subject to our terms of use, privacy policy, advertisement policy. © 2020 Minerva Medical Treatment Pvt Ltd

Our comments section is governed by our Comments Policy . By posting comments at Medical Dialogues you automatically agree with our Comments Policy , Terms And Conditions and Privacy Policy .

Similar News