More Like this

1. Ensemble Machine Learning for Alzheimer’s disease Classification from Retinal Vasculature

   Lin, Hely ; Fang, Ruogu ( October 2021 , Biomedical Engineering Society Annual Meeting)

   null (Ed.)

   Introduction: Alzheimer’s disease (AD) causes progressive irreversible cognitive decline and is the leading cause of dementia. Therefore, a timely diagnosis is imperative to maximize neurological preservation. However, current treatments are either too costly or limited in availability. In this project, we explored using retinal vasculature as a potential biomarker for early AD diagnosis. This project focuses on stage 3 of a three-stage modular machine learning pipeline which consisted of image quality selection, vessel map generation, and classification [1]. The previous model only used support vector machine (SVM) to classify AD labels which limited its accuracy to 82%. In this project, random forest and gradient boosting were added and, along with SVM, combined into an ensemble classifier, raising the classification accuracy to 89%. Materials and Methods: Subjects classified as AD were those who were diagnosed with dementia in “Dementia Outcome: Alzheimer’s disease” from the UK Biobank Electronic Health Records. Five control groups were chosen with a 5:1 ratio of control to AD patients where the control patients had the same age, gender, and eye side image as the AD patient. In total, 122 vessel images from each group (AD and control) were used. The vessel maps were then segmented from fundus images through U-net. A t-test feature selection was first done on the training folds and the selected features was fed into the classifiers with a p-value threshold of 0.01. Next, 20 repetitions of 5-fold cross validation were performed where the hyperparameters were solely tuned on the training data. An ensemble classifier consisting of SVM, gradient boosting tree, and random forests was built and the final prediction was made through majority voting and evaluated on the test set. Results and Discussion: Through ensemble classification, accuracy increased by 4-12% relative to the individual classifiers, precision by 9-15%, sensitivity by 2-9%, specificity by at least 9-16%, and F1 score by 712%. Conclusions: Overall, a relatively high classification accuracy was achieved using machine learning ensemble classification with SVM, random forest, and gradient boosting. Although the results are very promising, a limitation of this study is that the requirement of needing images of sufficient quality decreased the amount of control parameters that can be implemented. However, through retinal vasculature analysis, this project shows machine learning’s high potential to be an efficient, more cost-effective alternative to diagnosing Alzheimer’s disease. Clinical Application: Using machine learning for AD diagnosis through retinal images will make screening available for a broader population by being more accessible and cost-efficient. Mobile device based screening can also be enabled at primary screening in resource-deprived regions. It can provide a pathway for future understanding of the association between biomarkers in the eye and brain. 

   more » « less
2. Population Atlas Analysis of Emerging Brain Structural Connections in the Human Fetus

   https://doi.org/10.1002/jmri.29057  

   Calixto, Camilo ; Machado‐Rivas, Fedel ; Karimi, Davood ; Velasco‐Annis, Clemente ; Cortes‐Albornoz, Maria Camila ; Afacan, Onur ; Warfield, Simon K. ; Gholipour, Ali ; Jaimes, Camilo ( October 2023 , Journal of Magnetic Resonance Imaging)

   A lack of in utero imaging data hampers our understanding of the connections in the human fetal brain. Generalizing observations from postmortem subjects and premature newborns is inaccurate due to technical and biological differences.

   To evaluate changes in fetal brain structural connectivity between 23 and 35 weeks postconceptional age using a spatiotemporal atlas of diffusion tensor imaging (DTI).

   Retrospective.

   Publicly available diffusion atlases, based on 60 healthy women (age 18–45 years) with normal prenatal care, from 23 and 35 weeks of gestation.

   3.0 Tesla/DTI acquired with diffusion‐weighted echo planar imaging (EPI).

   We performed whole‐brain fiber tractography from DTI images. The cortical plate of each diffusion atlas was segmented and parcellated into 78 regions derived from the Edinburgh Neonatal Atlas (ENA33). Connectivity matrices were computed, representing normalized fiber connections between nodes. We examined the relationship between global efficiency (GE), local efficiency (LE), small‐worldness (SW), nodal efficiency (NE), and betweenness centrality (BC) with gestational age (GA) and with laterality.

   Linear regression was used to analyze changes in GE, LE, NE, and BC throughout gestation, and to assess changes in laterality. Thet‐tests were used to assess SW.P‐values were corrected using Holm‐Bonferroni method. A correctedP‐value <0.05 was considered statistically significant.

   Network analysis revealed a significant weekly increase in GE (5.83%/week, 95% CI 4.32–7.37), LE (5.43%/week, 95% CI 3.63–7.25), and presence of SW across GA. No significant hemisphere differences were found in GE (P = 0.971) or LE (P = 0.458). Increasing GA was significantly associated with increasing NE in 41 nodes, increasing BC in 3 nodes, and decreasing BC in 2 nodes.

   Extensive network development and refinement occur in the second and third trimesters, marked by a rapid increase in global integration and local segregation.

   3

   Stage 2

    

   more » « less
3. Revisiting the Trustworthiness of Saliency Methods in Radiology AI

   https://doi.org/10.1148/ryai.220221  

   Zhang, Jiajin ; Chao, Hanqing ; Dasegowda, Giridhar ; Wang, Ge ; Kalra, Mannudeep K. ; Yan, Pingkun ( January 2024 , Radiology: Artificial Intelligence)

   Purpose: To determine if saliency maps in radiology artificial intelligence (AI) are vulnerable to subtle perturbations of the input, which could potentially lead to misleading interpretations, using Prediction-Saliency Correlation (PSC) for evaluating the sensitivity and robustness of saliency methods. Materials and Methods: In this retrospective study, locally trained deep learning models and a research prototype provided by a commercial vender were systematically evaluated on 191,229 chest radiographs from the CheXpert dataset(1,2) and 7,022 MRI images of human brain tumor classification dataset(3). Two radiologists performed a reader study on 270 chest radiographs pairs. A model-agnostic approach for computing the PSC coefficient was used to evaluate the sensitivity and robustness of seven commonly used saliency methods. Results: Leveraging locally trained model parameters, we revealed the saliency methods’ low sensitivity (maximum PSC = 0.25, 95% CI: 0.12, 0.38) and weak robustness (maximum PSC = 0.12, 95% CI: 0.0, 0.25) on the CheXpert dataset. Without model specifics, we also showed that the saliency maps from a commercial prototype could be irrelevant to the model output (area under the receiver operating characteristic curve dropped by 8.6% without affecting the saliency map). The human observer studies confirmed that is difficult for experts to identify the perturbed images, who had less than 44.8% correctness. Conclusion: Popular saliency methods scored low PSC values on the two datasets of perturbed chest radiographs, indicating weak sensitivity and robustness. The proposed PSC metric provides a valuable quantification tool for validating the trustworthiness of medical AI explainability. Abbreviations: AI = artificial intelligence, PSC = prediction-saliency correlation, AUC = area under the receiver operating characteristic curve, SSIM = structural similarity index measure. Summary: Systematic evaluation of saliency methods through subtle perturbations in chest radiographs and brain MRI images demonstrated low sensitivity and robustness of those methods, warranting caution when using saliency methods that may misrepresent changes in AI model prediction. 

   more » « less
4. An MRI study of the corpus callosum in monkeys: Developmental trajectories and effects of neonatal hippocampal and amygdala lesions

   https://doi.org/10.1002/dev.21514  

   Payne, Christa ; Cirilli, Laetitia ; Bachevalier, Jocelyne ( April 2017 , Developmental Psychobiology)

   This study provides the first characterization of early developmental trajectories of corpus callosum (CC) segments in rhesus macaques using noninvasive MRI techniques and assesses long‐term effects of neonatal amygdala or hippocampal lesions on CC morphometry. In Experiment 1, 10 monkeys (5 males) were scanned at 1 week—2 years of age; eight additional infants (4 males) were scanned once at 1–4 weeks of age. The first 8 months showed marked growth across all segments, with sustained, albeit slower, growth through 24 months. Males and females had comparable patterns of CC maturation overall, but exhibited slight differences in the anterior and posterior segments, with greater increases in the isthmus for males and greater increases in the rostrum for females. The developmental changes are likely a consequence of varying degrees of axonal myelination, redirection, and pruning. In Experiment 2, animals with neonatal lesions of the amygdala (n = 6; 3 males) or hippocampus (n = 6; 4 males) were scanned at 1.5 years post‐surgery and compared to scans of six control animals from Experiment 1. Whereas amygdala damage yielded larger rostral and posterior body segments, hippocampal damage yielded larger rostrum and isthmus. These differences demonstrate that early perturbations to one medial temporal lobe structure may produce extensive and long‐lasting repercussions in other brain areas. The current findings emphasize the complexity of neural circuitry putatively subserving neurodevelopmental disorders such as autism spectrum disorder and Williams syndrome, which are each characterized by malformations and dysfunction of complex neural networks that include regions of the medial temporal lobe.

    

   more » « less
5. Metabolic alterations caused by the mutation and overexpression of the Tmem135 gene

   https://doi.org/10.1177/1535370220932856  

   Lee, Wei-Hua ; Bhute, Vijesh J ; Higuchi, Hitoshi ; Ikeda, Sakae ; Palecek, Sean P ; Ikeda, Akihiro ( November 2020 , Experimental Biology and Medicine)

   null (Ed.)

   Mitochondria are dynamic organelles that undergo fission and fusion. While they are essential for cellular metabolism, the effect of dysregulated mitochondrial dynamics on cellular metabolism is not fully understood. We previously found that transmembrane protein 135 ( Tmem135) plays a role in the regulation of mitochondrial dynamics in mice. Mice homozygous for a Tmem135 mutation ( Tmem135 FUN025/FUN025 ) display accelerated aging and age-related disease pathologies in the retina including the retinal pigment epithelium (RPE). We also generated a transgenic mouse line globally overexpressing the Tmem135 gene ( Tmem135 TG). In several tissues and cells that we studied such as the retina, heart, and fibroblast cells, we observed that the Tmem135 mutation causes elongated mitochondria, while overexpression of Tmem135 results in fragmented mitochondria. To investigate how abnormal mitochondrial dynamics affect metabolic signatures of tissues and cells, we identified metabolic changes in primary RPE cell cultures as well as heart, cerebellum, and hippocampus isolated from Tmem135 FUN025/FUN025 mice (fusion > fission) and Tmem135 TG mice (fusion < fission) using nuclear magnetic resonance spectroscopy. Metabolomics analysis revealed a tissue-dependent response to Tmem135 alterations, whereby significant metabolic changes were observed in the heart of both Tmem135 mutant and TG mice as compared to wild-type, while negligible effects were observed in the cerebellum and hippocampus. We also observed changes in Tmem135 FUN025/FUN025 and Tmem135 TG RPE cells associated with osmosis and glucose and phospholipid metabolism. We observed depletion of NAD + in both Tmem135 FUN025/FUN025 and Tmem135 TG RPE cells, indicating that imbalance in mitochondrial dynamics to both directions lowers the cellular NAD + level. Metabolic changes identified in this study might be associated with imbalanced mitochondrial dynamics in heart tissue and RPE cells which can likely lead to functional abnormalities. Impact statement Mitochondria are dynamic organelles undergoing fission and fusion. Proper regulation of this process is important for healthy aging process, as aberrant mitochondrial dynamics are associated with several age-related diseases/pathologies. However, it is not well understood how imbalanced mitochondrial dynamics may lead to those diseases and pathologies. Here, we aimed to determine metabolic alterations in tissues and cells from mouse models with over-fused (fusion > fission) and over-fragmented (fusion < fission) mitochondria that display age-related disease pathologies. Our results indicated tissue-dependent sensitivity to these mitochondrial changes, and metabolic pathways likely affected by aberrant mitochondrial dynamics. This study provides new insights into how dysregulated mitochondrial dynamics could lead to functional abnormalities of tissues and cells. 

   more » « less