SLIViT: a general AI framework for clinical-feature diagnosis from limited 3D biomedical-imaging data
Abstract We present SLIViT, a deep-learning framework that accurately measures disease-related risk factors in volumetric biomedical imaging, such as magnetic resonance imaging (MRI) scans, optical coherence tomography (OCT) scans, and ultrasound videos. To evaluate SLIViT, we applied it to five different datasets of these three different data modalities tackling seven learning tasks (including both classification and regression) and found that it consistently and significantly outperforms domain-specific state-of-the-art models, typically improving performance (ROC AUC or correlation) by 0.1-0.4. Notably, compared to existing approaches, SLIViT can be applied even when only a small number of annotated training samples is available, which is often a constraint in medical applications. When trained on less than 700 annotated volumes, SLIViT obtained accuracy comparable to trained clinical specialists while reducing annotation time by a factor of 5,000 demonstrating its utility to automate and expedite ongoing research and other practical clinical scenarios. *Oren Avram and Berkin Durmus equally contributed to this work. **Srinivas R. Sadda and Eran Halperin jointly supervised this study..
| Medienart: |
|
|---|
Preprint| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
ResearchSquare.com - (2023) vom: 21. Nov. Zur Gesamtaufnahme - year:2023 |
|---|
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Avram*, Oren [VerfasserIn] |
|---|
| Links: |
Volltext [kostenfrei] |
|---|
| Themen: |
|---|
| doi: |
10.21203/rs.3.rs-3044914/v2 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
XRA041606396 |
|---|
| LEADER | 01000caa a22002652 4500 | ||
|---|---|---|---|
| 001 | XRA041606396 | ||
| 003 | DE-627 | ||
| 005 | 20231205143642.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231121s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.21203/rs.3.rs-3044914/v2 |2 doi | |
| 035 | |a (DE-627)XRA041606396 | ||
| 035 | |a (ResearchSquare)10.21203/rs.3.rs-3044914/v2 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Avram*, Oren |e verfasserin |0 (orcid)0000-0003-1984-2139 |4 aut | |
| 245 | 1 | 0 | |a SLIViT: a general AI framework for clinical-feature diagnosis from limited 3D biomedical-imaging data |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a Computermedien |b c |2 rdamedia | ||
| 338 | |a Online-Ressource |b cr |2 rdacarrier | ||
| 520 | |a Abstract We present SLIViT, a deep-learning framework that accurately measures disease-related risk factors in volumetric biomedical imaging, such as magnetic resonance imaging (MRI) scans, optical coherence tomography (OCT) scans, and ultrasound videos. To evaluate SLIViT, we applied it to five different datasets of these three different data modalities tackling seven learning tasks (including both classification and regression) and found that it consistently and significantly outperforms domain-specific state-of-the-art models, typically improving performance (ROC AUC or correlation) by 0.1-0.4. Notably, compared to existing approaches, SLIViT can be applied even when only a small number of annotated training samples is available, which is often a constraint in medical applications. When trained on less than 700 annotated volumes, SLIViT obtained accuracy comparable to trained clinical specialists while reducing annotation time by a factor of 5,000 demonstrating its utility to automate and expedite ongoing research and other practical clinical scenarios. *Oren Avram and Berkin Durmus equally contributed to this work. **Srinivas R. Sadda and Eran Halperin jointly supervised this study. | ||
| 650 | 4 | |a Biology |7 (dpeaa)DE-84 | |
| 650 | 4 | |a 570 |7 (dpeaa)DE-84 | |
| 700 | 1 | |a Durmus*, Berkin |4 aut | |
| 700 | 1 | |a Rakocz, Nadav |4 aut | |
| 700 | 1 | |a Corradetti, Giulia |4 aut | |
| 700 | 1 | |a An, Ulzee |4 aut | |
| 700 | 1 | |a Nitalla, Muneeswar G. |4 aut | |
| 700 | 1 | |a Rudas, Ákos |0 (orcid)0000-0003-4346-8239 |4 aut | |
| 700 | 1 | |a Wakatsuki, Yu |4 aut | |
| 700 | 1 | |a Hirabayashi, Kazutaka |4 aut | |
| 700 | 1 | |a Velaga, Swetha |4 aut | |
| 700 | 1 | |a Tiosano, Liran |4 aut | |
| 700 | 1 | |a Corvi, Federico |0 (orcid)0000-0002-2661-5500 |4 aut | |
| 700 | 1 | |a Verma, Aditya |4 aut | |
| 700 | 1 | |a Karamat, Ayesha |4 aut | |
| 700 | 1 | |a Lindenberg, Sophiana |4 aut | |
| 700 | 1 | |a Oncel, Deniz |4 aut | |
| 700 | 1 | |a Almidani, Louay |4 aut | |
| 700 | 1 | |a Hull, Victoria |4 aut | |
| 700 | 1 | |a Fasih-Ahmad, Sohaib |4 aut | |
| 700 | 1 | |a Esmaeilkhanian, Houri |4 aut | |
| 700 | 1 | |a Wykoff, Charles C. |4 aut | |
| 700 | 1 | |a Rahmani, Elior |4 aut | |
| 700 | 1 | |a Arnold, Corey W. |4 aut | |
| 700 | 1 | |a Zhou, Bolei |4 aut | |
| 700 | 1 | |a Zaitlen, Noah |4 aut | |
| 700 | 1 | |a Gronau, Ilan |4 aut | |
| 700 | 1 | |a Sankararaman, Sriram |4 aut | |
| 700 | 1 | |a Chiang, Jeffrey N. |0 (orcid)0000-0002-6843-1355 |4 aut | |
| 700 | 1 | |a Sadda**, Srinivas R. |4 aut | |
| 700 | 1 | |a Halperin**, Eran |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t ResearchSquare.com |g (2023) vom: 21. Nov. |
| 773 | 1 | 8 | |g year:2023 |g day:21 |g month:11 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.21203/rs.3.rs-3044914/v2 |z kostenfrei |3 Volltext |
| 912 | |a GBV_XRA | ||
| 951 | |a AR | ||
| 952 | |j 2023 |b 21 |c 11 | ||