Akihiko Minami
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Akihiko Minami has a total of 30 co-authors. Here are the top co-authors by number of shared publications:
- Mami Iima
Department of Fundamental Development for Advanced Low Invasive Diagnostic Imaging, Nagoya University Graduate School of Medicine, Nagoya, Japan
2 shared publications - Masako Kataoka
Preemptive Medicine and Lifestyle-related Disease Research Center, Kyoto University Hospital, Kyoto, Japan
2 shared publications Akiyo Shinde
Department of Neurology, Tenri Hospital
1 shared publication- Ayako Ono
Department of Radiology, Tenri Hospital
1 shared publication - Gosuke Okubo
Department of Radiology, Tenri Hospital
1 shared publication - Hyunjin Kim
Department of Radiology, Tenri Hospital, Nara, Japan
1 shared publication
Latest Publications
- 1 January 2025
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 7 (1) , e240206
Deep Learning Applied to Diffusion-weighted Imaging for Differentiating Malignant from Benign Breast Tumors without Lesion Segmentation
Deep learning models, particularly small two-dimensional convolutional neural networks, showed good performance in differentiating between benign and malignant breast tumors using diffusion-weighted MRI, compared with breast radiologists.
- 1 January 2025
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 7 (1) , e240206
Deep Learning Applied to Diffusion-weighted Imaging for Differentiating Malignant from Benign Breast Tumors without Lesion Segmentation
Deep learning models, particularly small two-dimensional convolutional neural networks, showed good performance in differentiating between benign and malignant breast tumors using diffusion-weighted MRI, compared with breast radiologists.
- 25 December 2024
- Published by Tenri Foundation, Tenri Institute of Medical Research in Tenri Medical Bulletin
- Vol. 27 (1) , 33-39
Adult-onset cerebello-brainstem-dominant form of X-linked adrenoleukodystrophy with auditory pathway involvement: A case report
X-linked adrenoleukodystrophy (X-ALD) is classified into several clinical subtypes. Among them, the cerebello-brainstem-dominant form is rare, and its symptoms and imaging findings can mimic spinocerebellar degeneration (SCD), making diagnosis difficult. Herein, we report a case that we initially diagnosed as SCD but was eventually diagnosed as the cerebello-brainstem-dominant form of X-ALD. The patient presented with psychiatric symptoms in adulthood. He subsequently developed gait disturbance and was referred to the neurology department of our hospital. Initial magnetic resonance imaging (MRI) showed atrophy of the cerebellum and pons, and abnormal signals in various regions, including the middle cerebellar peduncles, the corpus callosum, infero-posterior part of the bilateral thalami, and lateral edges of the midbrain, on T2-weighted imaging. The patient was followed up as SCD; however, his symptoms and MRI findings worsened. Further examination was performed, and the diagnosis of X-ALD was made. Abnormal signals in the thalami and midbrain on MRI are considered to correspond to the medial geniculate bodies and the brachia of the inferior colliculus, respectively. They constitute the auditory pathway, which is an evaluation item in Loes score, an imaging-based scoring system for patients with X-ALD. Recognizing abnormalities in the bilateral auditory pathway may aid accurate diagnosis of X-ALD.
- 25 December 2024
- Published by Tenri Foundation, Tenri Institute of Medical Research in Tenri Medical Bulletin
- Vol. 27 (1) , 33-39
Adult-onset cerebello-brainstem-dominant form of X-linked adrenoleukodystrophy with auditory pathway involvement: A case report
X-linked adrenoleukodystrophy (X-ALD) is classified into several clinical subtypes. Among them, the cerebello-brainstem-dominant form is rare, and its symptoms and imaging findings can mimic spinocerebellar degeneration (SCD), making diagnosis difficult. Herein, we report a case that we initially diagnosed as SCD but was eventually diagnosed as the cerebello-brainstem-dominant form of X-ALD. The patient presented with psychiatric symptoms in adulthood. He subsequently developed gait disturbance and was referred to the neurology department of our hospital. Initial magnetic resonance imaging (MRI) showed atrophy of the cerebellum and pons, and abnormal signals in various regions, including the middle cerebellar peduncles, the corpus callosum, infero-posterior part of the bilateral thalami, and lateral edges of the midbrain, on T2-weighted imaging. The patient was followed up as SCD; however, his symptoms and MRI findings worsened. Further examination was performed, and the diagnosis of X-ALD was made. Abnormal signals in the thalami and midbrain on MRI are considered to correspond to the medial geniculate bodies and the brachia of the inferior colliculus, respectively. They constitute the auditory pathway, which is an evaluation item in Loes score, an imaging-based scoring system for patients with X-ALD. Recognizing abnormalities in the bilateral auditory pathway may aid accurate diagnosis of X-ALD.
- 17 February 2024
- Published by Wiley in Journal of Magnetic Resonance Imaging
- Vol. 60 (5) , 2237-2238
Comments on “A Channel‐Dimensional Feature‐Reconstructed Deep Learning Model for Predicting Breast Cancer Molecular Subtypes on Overall b‐Value Diffusion‐Weighted MRI”
- 17 February 2024
- Published by Wiley in Journal of Magnetic Resonance Imaging
- Vol. 60 (5) , 2237-2238
Comments on “A Channel‐Dimensional Feature‐Reconstructed Deep Learning Model for Predicting Breast Cancer Molecular Subtypes on Overall b‐Value Diffusion‐Weighted MRI”
- Open Access
- 25 December 2023
- Published by Tenri Foundation, Tenri Institute of Medical Research in Tenri Medical Bulletin
- Vol. 26 (2) , 120-122
Differentiation between Staphylococcus aureus and coagulase-negative staphylococci using deep learning
【はじめに】血液培養陽性例において,病原性の強い黄色ブドウ球菌と病原性の低いコアグラーゼ陰性ブドウ球菌(CNS)の迅速な判別が求められるが,グラム染色による形態的な区別には困難を伴う.そこで本研究では人工知能による分類アルゴリズムを導入し,鏡検画像を分類することを目的とした.【方法】2019 年から2020 年の入院患者の好気性ボトルにおける血液培養陽性症例のうち,黄色ブドウ球菌とCNS のグラム染色検体の撮影を行い,学習・検証用データと感度・特異度を評価するための診断精度評価用データに振り分けた.学習・検証用データは512 × 512 ピクセルの画像に分割し,それを手作業で黄色ブドウ球菌,CNS,背景の3 つに分類した.Python+Tensorflow を用いてInceptionResNet V2 による転移学習アルゴリズムを組み込んだ畳み込みニューラルネットワークを構築し,Amazon Web Service 上にて教師あり分類として手作業で分類した学習・検証用データを学習させた.また,この学習済みモデルと診断精度評価用データを用いて以下のように感度・特異度の評価を行った.鏡検画像を512 × 512 ピクセルのセルに分割し,それぞれのセルの黄色ブドウ球菌・CNS・背景の確率を出力し,黄色ブドウ球菌またはCNS と判断されたセルの黄色ブドウ球菌の確率の平均をその鏡検画像の黄色ブドウ球菌「らしさ」として用いた.このスコアを用いてMALDI-TOF の結果をゴールドスタンダードとした感度・特異度を評価した.【結果】黄色ブドウ球菌32 症例とCNS の症例15 症例のグラム染色像を撮影し,それぞれ22 症例と10 症例を学習・検証用データとして,10 症例と5 症例を診断精度評価用データとして用いた.学習・検証用データの画像を512 × 512 ピクセルの画像に分割した結果,黄色ブドウ球菌の画像5,163 枚,CNS の画像6,313 枚,背景の画像13,955 枚 を得た.これの7 割17,804 枚を学習用に,3 割7,628 枚を検証用に用いた.これにより学習を行い,診断精度評価用データにて感度・特異度評価を行ったところ,感度・特異度はそれぞれ67% と87%(AUC 0.778)となった.
- 25 December 2023
- Published by Tenri Foundation, Tenri Institute of Medical Research in Tenri Medical Bulletin
- Vol. 26 (2) , 120-122
Open AccessDifferentiation between Staphylococcus aureus and coagulase-negative staphylococci using deep learning
【はじめに】血液培養陽性例において,病原性の強い黄色ブドウ球菌と病原性の低いコアグラーゼ陰性ブドウ球菌(CNS)の迅速な判別が求められるが,グラム染色による形態的な区別には困難を伴う.そこで本研究では人工知能による分類アルゴリズムを導入し,鏡検画像を分類することを目的とした.【方法】2019 年から2020 年の入院患者の好気性ボトルにおける血液培養陽性症例のうち,黄色ブドウ球菌とCNS のグラム染色検体の撮影を行い,学習・検証用データと感度・特異度を評価するための診断精度評価用データに振り分けた.学習・検証用データは512 × 512 ピクセルの画像に分割し,それを手作業で黄色ブドウ球菌,CNS,背景の3 つに分類した.Python+Tensorflow を用いてInceptionResNet V2 による転移学習アルゴリズムを組み込んだ畳み込みニューラルネットワークを構築し,Amazon Web Service 上にて教師あり分類として手作業で分類した学習・検証用データを学習させた.また,この学習済みモデルと診断精度評価用データを用いて以下のように感度・特異度の評価を行った.鏡検画像を512 × 512 ピクセルのセルに分割し,それぞれのセルの黄色ブドウ球菌・CNS・背景の確率を出力し,黄色ブドウ球菌またはCNS と判断されたセルの黄色ブドウ球菌の確率の平均をその鏡検画像の黄色ブドウ球菌「らしさ」として用いた.このスコアを用いてMALDI-TOF の結果をゴールドスタンダードとした感度・特異度を評価した.【結果】黄色ブドウ球菌32 症例とCNS の症例15 症例のグラム染色像を撮影し,それぞれ22 症例と10 症例を学習・検証用データとして,10 症例と5 症例を診断精度評価用データとして用いた.学習・検証用データの画像を512 × 512 ピクセルの画像に分割した結果,黄色ブドウ球菌の画像5,163 枚,CNS の画像6,313 枚,背景の画像13,955 枚 を得た.これの7 割17,804 枚を学習用に,3 割7,628 枚を検証用に用いた.これにより学習を行い,診断精度評価用データにて感度・特異度評価を行ったところ,感度・特異度はそれぞれ67% と87%(AUC 0.778)となった.