Improving performance robustness of subject-based brain segmentation software
Purpose Artificial intelligence (AI)-based image analysis tools to quantify the brain have become commercialized. However, insufficient data for learning and scanner specificity is a limitation for achieving high quality. In the present study, the performance of personalized brain segmentation softw...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Korean Encephalitis and Neuroinflammation Society
2023-01-01
|
| Series: | Encephalitis |
| Subjects: | |
| Online Access: | http://www.encephalitisjournal.org/upload/pdf/encephalitis-2022-00108.pdf |
| _version_ | 1797957210825818112 |
|---|---|
| author | Jong-Hyeok Park Kyung-Il Park Dongmin Kim Myungjae Lee Shinuk Kang Seung Joo Kang Dae Hyun Yoon |
| author_facet | Jong-Hyeok Park Kyung-Il Park Dongmin Kim Myungjae Lee Shinuk Kang Seung Joo Kang Dae Hyun Yoon |
| author_sort | Jong-Hyeok Park |
| collection | DOAJ |
| description | Purpose Artificial intelligence (AI)-based image analysis tools to quantify the brain have become commercialized. However, insufficient data for learning and scanner specificity is a limitation for achieving high quality. In the present study, the performance of personalized brain segmentation software when applied to multicenter data using an AI model trained on data from a single institution was improved. Methods Preindicators of brain white matter (WM) information from the training dataset were utilized for preprocessing. During learning, data of cognitively normal (CN) individuals from a single center were utilized, and data of CN individuals and Alzheimer disease (AD) patients enrolled in multiple centers were considered the test set. Results The preprocessing based on the preindicator (dice similarity coefficient [DSC], 0.8567) resulted in a better performance than without (DSC, 0.7921). The standard deviation (SD) of the WM region intensity (DSC, 0.8303) had a more substantial influence on the performance than the average intensity (DSC, 0.6591). When the SD of the test data WM intensity was smaller than the learning data, the performance improved (0.03 increase in lower SD, 0.05 decrease in higher SD). Furthermore, preindicator-based pretreatment increased the correlation of mean cortical thickness of the entire gray matter between Atroscan and FreeSurfer, and data augmentation without preprocessing did not.Both preindicator processing and data augmentation improved the correlation coefficient from 0.7584 to 0.8165. Conclusion Data augmentation and preindicator-based preprocessing of training data can improve the performance of AI-based brain segmentation software, both increasing the generalizability and stability of brain segmentation software. |
| first_indexed | 2024-04-11T00:00:47Z |
| format | Article |
| id | doaj.art-203d451a371b43098fe5d14815865214 |
| institution | Directory Open Access Journal |
| issn | 2765-4559 2734-1461 |
| language | English |
| last_indexed | 2024-04-11T00:00:47Z |
| publishDate | 2023-01-01 |
| publisher | Korean Encephalitis and Neuroinflammation Society |
| record_format | Article |
| series | Encephalitis |
| spelling | doaj.art-203d451a371b43098fe5d148158652142023-01-10T00:47:56ZengKorean Encephalitis and Neuroinflammation SocietyEncephalitis2765-45592734-14612023-01-0131243310.47936/encephalitis.2022.0010846Improving performance robustness of subject-based brain segmentation softwareJong-Hyeok Park0Kyung-Il Park1Dongmin Kim2Myungjae Lee3Shinuk Kang4Seung Joo Kang5Dae Hyun Yoon6 JLK, Seoul, Korea Department of Neurology, Seoul National University Hospital, Seoul National University College of Medicine, Seoul, Korea JLK, Seoul, Korea JLK, Seoul, Korea JLK, Seoul, Korea Division of Gastroenterology, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Korea Division of Psychiatry, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, KoreaPurpose Artificial intelligence (AI)-based image analysis tools to quantify the brain have become commercialized. However, insufficient data for learning and scanner specificity is a limitation for achieving high quality. In the present study, the performance of personalized brain segmentation software when applied to multicenter data using an AI model trained on data from a single institution was improved. Methods Preindicators of brain white matter (WM) information from the training dataset were utilized for preprocessing. During learning, data of cognitively normal (CN) individuals from a single center were utilized, and data of CN individuals and Alzheimer disease (AD) patients enrolled in multiple centers were considered the test set. Results The preprocessing based on the preindicator (dice similarity coefficient [DSC], 0.8567) resulted in a better performance than without (DSC, 0.7921). The standard deviation (SD) of the WM region intensity (DSC, 0.8303) had a more substantial influence on the performance than the average intensity (DSC, 0.6591). When the SD of the test data WM intensity was smaller than the learning data, the performance improved (0.03 increase in lower SD, 0.05 decrease in higher SD). Furthermore, preindicator-based pretreatment increased the correlation of mean cortical thickness of the entire gray matter between Atroscan and FreeSurfer, and data augmentation without preprocessing did not.Both preindicator processing and data augmentation improved the correlation coefficient from 0.7584 to 0.8165. Conclusion Data augmentation and preindicator-based preprocessing of training data can improve the performance of AI-based brain segmentation software, both increasing the generalizability and stability of brain segmentation software.http://www.encephalitisjournal.org/upload/pdf/encephalitis-2022-00108.pdfalzheimer diseaseartificial intelligencedata augmentationsegmentation |
| spellingShingle | Jong-Hyeok Park Kyung-Il Park Dongmin Kim Myungjae Lee Shinuk Kang Seung Joo Kang Dae Hyun Yoon Improving performance robustness of subject-based brain segmentation software Encephalitis alzheimer disease artificial intelligence data augmentation segmentation |
| title | Improving performance robustness of subject-based brain segmentation software |
| title_full | Improving performance robustness of subject-based brain segmentation software |
| title_fullStr | Improving performance robustness of subject-based brain segmentation software |
| title_full_unstemmed | Improving performance robustness of subject-based brain segmentation software |
| title_short | Improving performance robustness of subject-based brain segmentation software |
| title_sort | improving performance robustness of subject based brain segmentation software |
| topic | alzheimer disease artificial intelligence data augmentation segmentation |
| url | http://www.encephalitisjournal.org/upload/pdf/encephalitis-2022-00108.pdf |
| work_keys_str_mv | AT jonghyeokpark improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT kyungilpark improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT dongminkim improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT myungjaelee improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT shinukkang improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT seungjookang improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware AT daehyunyoon improvingperformancerobustnessofsubjectbasedbrainsegmentationsoftware |