MRI radiomics-based interpretable model and nomogram for preoperative prediction of Ki-67 expression status in primary central nervous system lymphoma DOI Creative Commons

Endong Zhao,

Yunfeng Yang,

Miaomiao Bai

и другие.

Frontiers in Medicine, Год журнала: 2024, Номер 11

Опубликована: Июнь 27, 2024

Objectives To investigate the value of interpretable machine learning model and nomogram based on clinical factors, MRI imaging features, radiomic features to predict Ki-67 expression in primary central nervous system lymphomas (PCNSL). Materials methods images information 92 PCNSL patients were retrospectively collected, which divided into 53 cases training set 39 external validation according different medical centers. A 3D brain tumor segmentation was trained nnU-NetV2, two prediction models, Random Forest (RF) incorporating SHapley Additive exPlanations (SHAP) method multivariate logistic regression, proposed for task status prediction. Results The mean dice Similarity Coefficient (DSC) score 0.85. On task, AUC RF 0.84 (95% CI:0.81, 0.86; p < 0.001), a 3% improvement compared nomogram. Delong test showed that z statistic difference between models 1.901, corresponding 0.057. In addition, SHAP analysis Rad-Score made significant contribution decision. Conclusion this study, we developed used an preoperative patients, improved task. Clinical relevance statement represents degree active cell proliferation is important prognostic parameter associated with outcomes. Non-invasive accurate level preoperatively plays role targeting treatment selection patient stratification management thereby improving prognosis.

Язык: Английский

Deep learning for medical image segmentation: State-of-the-art advancements and challenges DOI Creative Commons
Md. Eshmam Rayed,

S. M. Sajibul Islam,

Sadia Islam Niha

и другие.

Informatics in Medicine Unlocked, Год журнала: 2024, Номер 47, С. 101504 - 101504

Опубликована: Янв. 1, 2024

Image segmentation, a crucial process of dividing images into distinct parts or objects, has witnessed remarkable advancements with the emergence deep learning (DL) techniques. The use layers in neural networks, like object form recognition higher and basic edge identification lower layers, markedly improved quality accuracy image segmentation. Consequently, DL using picture segmentation become commonplace, video analysis, facial recognition, etc. Grasping applications, algorithms, current performance, challenges are for advancing DL-based medical However, there's lack studies delving latest state-of-the-art developments this field. Therefore, survey aimed to thoroughly explore most recent applications encompassing an in-depth analysis various commonly used datasets, pre-processing techniques algorithms. This study also investigated advancement done by analyzing their results experimental details. Finally, discussed future research directions Overall, provides comprehensive insight covering its application domains, model exploration, results, challenges, directions—a valuable resource multidisciplinary studies.

Язык: Английский

Процитировано

44
Enhancing medical image segmentation with MA-UNet: a multi-scale attention framework DOI
Hongzhi Li, Z. Y. Ren,

Zhu Guoqing

и другие.

The Visual Computer, Год журнала: 2025, Номер unknown

Опубликована: Янв. 5, 2025

Язык: Английский

Процитировано

3
A review of deep learning segmentation methods for carotid artery ultrasound images DOI
Qinghua Huang, Haozhe Tian,

Lizhi Jia

и другие.

Neurocomputing, Год журнала: 2023, Номер 545, С. 126298 - 126298

Опубликована: Май 5, 2023

Язык: Английский

Процитировано

40
MFFSP: Multi-scale feature fusion scene parsing network for landslides detection based on high-resolution satellite images DOI
Penglei Li, Yi Wang, Tongzhen Si

и другие.

Engineering Applications of Artificial Intelligence, Год журнала: 2023, Номер 127, С. 107337 - 107337

Опубликована: Окт. 27, 2023

Язык: Английский

Процитировано

23
Role of Artificial Intelligence in Medical Image Analysis: A Review of Current Trends and Future Directions DOI
Xin Li, Lei Zhang, Jingsi Yang

и другие.

Journal of Medical and Biological Engineering, Год журнала: 2024, Номер 44(2), С. 231 - 243

Опубликована: Апрель 1, 2024

Язык: Английский

Процитировано

12
C2FTFNet: Coarse-to-fine transformer network for joint optic disc and cup segmentation DOI
Yugen Yi, Yan Jiang, Bin Zhou

и другие.

Computers in Biology and Medicine, Год журнала: 2023, Номер 164, С. 107215 - 107215

Опубликована: Июль 5, 2023

Язык: Английский

Процитировано

21
Clinical assessment of deep learning-based uncertainty maps in lung cancer segmentation DOI
Federica Carmen Maruccio, Wietse S.C. Eppinga, Max-Heinrich Laves

и другие.

Physics in Medicine and Biology, Год журнала: 2024, Номер 69(3), С. 035007 - 035007

Опубликована: Янв. 3, 2024

. Prior to radiation therapy planning, accurate delineation of gross tumour volume (GTVs) and organs at risk (OARs) is crucial. In the current clinical practice, performed manually by oncologists, which time-consuming prone large inter-observer variability. With advent deep learning (DL) models, automated contouring has become possible, speeding up procedures assisting clinicians. However, these tools are currently used in clinic mostly for OARs, since systems not reliable yet GTVs. To improve reliability systems, researchers have started exploring topic probabilistic neural networks. there still limited knowledge practical implementation such networks real settings.

Язык: Английский

Процитировано

8
Multi-scale coal and gangue detection in dense state based on improved Mask RCNN DOI
Xi Wang, Shuang Wang, Yongcun Guo

и другие.

Measurement, Год журнала: 2023, Номер 221, С. 113467 - 113467

Опубликована: Авг. 25, 2023

Язык: Английский

Процитировано

14
Artificial Intelligence in Obstetric Anomaly Scan: Heart and Brain DOI Creative Commons
Iuliana-Alina Enache,

Cătălina Iovoaica-Rămescu,

Ștefan Gabriel Ciobanu

и другие.

Life, Год журнала: 2024, Номер 14(2), С. 166 - 166

Опубликована: Янв. 23, 2024

Background: The ultrasound scan represents the first tool that obstetricians use in fetal evaluation, but sometimes, it can be limited by mobility or position, excessive thickness of maternal abdominal wall, presence post-surgical scars on wall. Artificial intelligence (AI) has already been effectively used to measure biometric parameters, automatically recognize standard planes and for disease diagnosis, which helps conventional imaging methods. usage information, images, a machine learning program create an algorithm capable assisting healthcare providers reducing workload, duration examination, increasing correct diagnosis capability. recent remarkable expansion electronic medical records diagnostic coincides with enormous success algorithms image identification tasks. Objectives: We aim review most relevant studies based deep anomaly evaluation complex systems (heart brain), enclose frequent anomalies.

Язык: Английский

Процитировано

5
CrossTransUnet: A New Computationally Inexpensive Tumor Segmentation Model for Brain MRI DOI Creative Commons
Andrés Anaya-Isaza, Leonel Mera-Jiménez, Álvaro Fernández-Quilez

и другие.

IEEE Access, Год журнала: 2023, Номер 11, С. 27066 - 27085

Опубликована: Янв. 1, 2023

Brain tumors are usually fatal diseases with low life expectancies due to the organs they affect, even if benign. Diagnosis and treatment of these challenging tasks, for experienced physicians experts, heterogeneity tumor cells. In recent years, advances in deep learning (DL) methods have been integrated aid diagnosis, detection, segmentation brain neoplasms. However, is a computationally expensive process, typically based on convolutional neural networks (CNNs) UNet framework. While has shown promising results, new models developments can be incorporated into conventional architecture improve performance. this research, we propose three new, inexpensive, inspired by Transformers. These designed 4-stage encoder-decoder structure implement our cross-attention model, along separable convolution layers, avoid loss dimensionality activation maps reduce computational cost while maintaining high The attention model different configurations modifying transition encoder, decoder blocks. proposed evaluated against classical network, showing that differences up an order magnitude number training parameters. Additionally, one outperforms UNet, achieving significantly less time Dice Similarity Coefficient (DSC) 94%, ensuring effectiveness segmentation.

Язык: Английский

Процитировано

11