Development of new diagnostic system for brain metastasis with deep learning

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K07645
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 52040:Radiological sciences-related
• Research Institution: Nagoya City University
• Principal Investigator: Hiwatashi Akio 名古屋市立大学, 医薬学総合研究院(医学), 教授 (30444855)
• Co-Investigator(Kenkyū-buntansha): 栂尾 理 九州大学, 医学研究院, 准教授 (10452749) ; 菊地 一史 九州大学, 医学研究院, 助教 (20529838) ; 石神 康生 九州大学, 医学研究院, 教授 (10403916)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 脳転移 / MRI / 機械学習

Outline of Final Research Achievements

VISIBLE (Volume Isotropic Simultaneous Interleaved Bright- and bLack-blood Examination) provides simultaneous acquisitions of images with blood vessel suppression (“Black images” hereafter) and without blood vessel suppression (“Bright images”) on postcontrast MR imaging. VISIBLE can reduce false-positives, such as those caused by insufficient suppression of blood vessels through the combined use of Bright images and Black images, however, the original version of this system about five minutes to obtain. In this research, we modified this sequence to obtain in three minutes. With this new sequence, we gathered patients data with and without metastases. Using deep learning, we created a new AI based diagnostic technique and examined its clinical utility.

Free Research Field

放射線診断学

Academic Significance and Societal Importance of the Research Achievements

脳転移は成人で最多の頭蓋内腫瘍であるが、正確な診断には時間を要し、見逃される病変が多い。VISIBLE (Volume Isotropic Simultaneous Interleaved Bright- and bLack-blood Examination)法は血管信号抑制画像と非抑制画像を同時取得し、高感度、低偽陽性率で脳転移を診断可能であるが、従来法では撮像に5分程度必要であった。そこで撮像時間の短縮を試みた。また、患者データベースを蓄積し、AIシステムを応用した脳転移の自動検出診断支援システムを作成し、放射線科医の診断にどの程度貢献できるかを評価した。