| Project/Area Number |
23K06454
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 49020:Human pathology-related
|
|---|
| Research Institution | Tokai University |
|---|
Principal Investigator |
カレーラス ジュアキム 東海大学, 医学部, 准教授 (90637191)
|
|---|
| Project Period (FY) |
2023-04-01 – 2026-03-31
|
| Project Status |
Granted (Fiscal Year 2024)
|
| Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2025: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2024: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
Fiscal Year 2023: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
|
|---|
| Keywords | immuno-oncology / non-Hodgkin lymphoma / artificial intelligence / deep learning / neural network / colitis / lymphoma / targeted sequencing / immune microenvironment / immune checkpoint / molecular pathology / hematological neoplasia / NGS / リンパ腫 / 免疫腫瘍学 / 人工知能 / 免疫チェックポイント / 次世代シーケンシング |
|---|
| Outline of Research at the Start |
この研究では、集学的アプローチを使用してリンパ腫の免疫チェックポイントを分析しています。B細胞性リンパ腫の解析を行います。重要なマーカーは、PD-L1、CSF1R、IL-10 などです。免疫組織化学、遺伝子発現解析、コピー数・LOH解析、次世代標的変異解析などを実施します。解析には、従来のバイオインフォマティクスだけでなく、人工知能（機械学習、ニューラルネットワーク）なども含まれます。
|
| Outline of Annual Research Achievements |
We have optimized the immunohistochemistry of immuno-oncology markers. The markers have been tested in non-Hodgkin lymphoma and in B-cell lymphomas such as FL, DLBCL, MALT, etc. Additionally, analysis have been performed in inflammatory bowel conditions such as ulcerative colitis, Chron's disease, and celiac disease. We have advanced in the image analysis classification part of the study, that is based on Artificial Intelligence (CNN and deep learning). Based on conventional H&E staining, we classified using RenNet-based architecture 550 lymphoma subtypes and 1,750,000 image patches.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The design of the convolutional neural network and optimization of the analysis took some time to design and implement. Several methods were tested. However, we managed to reach a deep learning performance of 99%, which is almost pathologist-human level. Therefore, now we are in position to accelerate the final analyses. Of note, we are experimenting with topology-based AI analysis, which is new type of analysis that we are testing in pilot samples of 150 lymphoma cases.
|
| Strategy for Future Research Activity |
The plan is the following: (1) Complete small and quick proof-of-concept projects that we publish in form of preprints and open access journals because we want to get quick feedback from other scientists and society; (2) Deep into the immunohistochemical-based analysis in more "conventional" pathology studies and journals; (3) Complete the deep learning AI-based image classification H&E lymphoma study; (4) Advance in the topology-based AI analysis.
|
