Frontiers in neurology · 2025
Comparison of Radiomics and conventional SUVr methods for Alzheimer's disease classification using AV45 PET imaging.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Gao Haiyan, Tan Arui, Wu Junhao, Cao Zhen, Zhu Ziyang, Zhang Wei
Аннотация
OBJECTIVE: To compare the diagnostic performance of radiomics-based analysis and the conventional standardized uptake value ratio (SUVr) method in classifying Alzheimer's disease (AD) and non-Alzheimer's disease (NAD) using AV45 PET imaging.
METHODS: This retrospective study included 79 patients diagnosed with AD and 34 patients diagnosed with NAD between July 2023 and August 2024. All patients underwent AV45 PET imaging, and the images were registered to a standard template for the extraction of SUVr metrics, including SUVmaxr, SUVmeanr, and SUVmoder, as well as radiomic features (a total of 660 features) from regions of interest (ROIs) in the brain lobes. Feature importance was ranked using a random forest algorithm, and three models were constructed: an SUVr model, a radiomics model, and a combined model. The classification performance was assessed using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA). Model accuracy, sensitivity, specificity, and precision were evaluated using the Mann-Whitney test, DeLong test, and confusion matrices.
RESULTS: There were no significant differences in gender and age between AD and NAD groups (p > 0.05). SUVr analysis showed no statistically significant differences in SUVmaxr values in the frontal and occipital lobes between AD and NAD patients, while SUVmeanr and SUVmoder in other lobes exhibited significant differences (p < 0.05). The 15 most important radiomic features were primarily concentrated in the temporal, frontal, and parietal lobes, with the highest-ranked features being original_firstorder_Skewness and original_glcm_ClusterShade. The area under the curve (AUC) of the Radiomics model was 0.89 (95% CI: 0.75-0.98), significantly higher than that of the SUVr model (AUC = 0.67, 95% CI: 0.45-0.86, p = 0.026). The combined model achieved an AUC of 0.88, showing no significant improvement over the Radiomics model alone. The Radiomics model outperformed the SUVr model in terms of accuracy (88% vs. 68%), sensitivity (96% vs. 78%), specificity (73% vs. 45%), and precision (88% vs. 75%). DCA analysis further confirmed the superior diagnostic performance of the Radiomics model.
CONCLUSION: The radiomics-based approach significantly outperformed the conventional SUVr method, particularly in terms of sensitivity and specificity. This study highlights the potential of radiomics for quantitative PET imaging analysis and its promising clinical applications.
METHODS: This retrospective study included 79 patients diagnosed with AD and 34 patients diagnosed with NAD between July 2023 and August 2024. All patients underwent AV45 PET imaging, and the images were registered to a standard template for the extraction of SUVr metrics, including SUVmaxr, SUVmeanr, and SUVmoder, as well as radiomic features (a total of 660 features) from regions of interest (ROIs) in the brain lobes. Feature importance was ranked using a random forest algorithm, and three models were constructed: an SUVr model, a radiomics model, and a combined model. The classification performance was assessed using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA). Model accuracy, sensitivity, specificity, and precision were evaluated using the Mann-Whitney test, DeLong test, and confusion matrices.
RESULTS: There were no significant differences in gender and age between AD and NAD groups (p > 0.05). SUVr analysis showed no statistically significant differences in SUVmaxr values in the frontal and occipital lobes between AD and NAD patients, while SUVmeanr and SUVmoder in other lobes exhibited significant differences (p < 0.05). The 15 most important radiomic features were primarily concentrated in the temporal, frontal, and parietal lobes, with the highest-ranked features being original_firstorder_Skewness and original_glcm_ClusterShade. The area under the curve (AUC) of the Radiomics model was 0.89 (95% CI: 0.75-0.98), significantly higher than that of the SUVr model (AUC = 0.67, 95% CI: 0.45-0.86, p = 0.026). The combined model achieved an AUC of 0.88, showing no significant improvement over the Radiomics model alone. The Radiomics model outperformed the SUVr model in terms of accuracy (88% vs. 68%), sensitivity (96% vs. 78%), specificity (73% vs. 45%), and precision (88% vs. 75%). DCA analysis further confirmed the superior diagnostic performance of the Radiomics model.
CONCLUSION: The radiomics-based approach significantly outperformed the conventional SUVr method, particularly in terms of sensitivity and specificity. This study highlights the potential of radiomics for quantitative PET imaging analysis and its promising clinical applications.