Wu, Teresa

There is a growing interest in the application of machine learning (ML) in Alzheimer’s disease (AD) research. However, neuropsychiatric symptoms (NPS), frequent in subjects with AD, mild cognitive impairment (MCI), and other related dementias have not been analyzed sufficiently using ML methods. To portray the landscape and potential of ML research in AD and NPS studies, we present a comprehensive literature review of existing ML approaches and commonly studied AD biomarkers. We conducted PubMed searches with keywords related to NPS, AD biomarkers, machine learning, and cognition. We included a total of 38 articles in this review after excluding some irrelevant studies from the search results and including 6 articles based on a snowball search from the bibliography of the relevant studies. We found a limited number of studies focused on NPS with or without AD biomarkers. In contrast, multiple statistical machine learning and deep learning methods have been used to build predictive diagnostic models using commonly known AD biomarkers. These mainly included multiple imaging biomarkers, cognitive scores, and various omics biomarkers. Deep learning approaches that combine these biomarkers or multi-modality datasets typically outperform single-modality datasets. We conclude ML may be leveraged to untangle the complex relationships of NPS and AD biomarkers with cognition. This may potentially help to predict the progression of MCI or dementia and develop more targeted early intervention approaches based on NPS.

Objectives The objective of this longitudinal study was to determine whether brain iron accumulation, measured using magnetic resonance imaging magnetic transverse relaxation rates (T2*), is associated with response to erenumab for the treatment of migraine. Methods Participants (n = 28) with migraine, diagnosed using international classification of headache disorders 3rd edition criteria, were eligible if they had six to 25 migraine days during a four-week headache diary run-in phase. Participants received two treatments with 140 mg erenumab, one immediately following the pre-treatment run-in phase and a second treatment four weeks later. T2* data were collected immediately following the pre-treatment phase, and at two weeks and eight weeks following the first erenumab treatment. Patients were classified as erenumab responders if their migraine-day frequency at five-to-eight weeks post-initial treatment was reduced by at least 50% compared to the pre-treatment run-in phase. A longitudinal Sandwich estimator approach was used to compare longitudinal group differences (responders vs non-responders) in T2* values, associated with iron accumulation. Group visit effects were calculated with a significance threshold of p = 0.005 and cluster forming threshold of 250 voxels. T2* values of 19 healthy controls were used for a reference. The average of each significant region was compared between groups and visits with Bonferroni corrections for multiple comparisons with significance defined as p < 0.05. Results Pre- and post-treatment longitudinal imaging data were available from 28 participants with migraine for a total of 79 quantitative T2* images. Average subject age was 42 ± 13 years (25 female, three male). Of the 28 subjects studied, 53.6% were erenumab responders. Comparing longitudinal T2* between erenumab responders vs non-responders yielded two comparisons which survived the significance threshold of p < 0.05 after correction for multiple comparisons: the difference at eight weeks between the erenumab-responders and non-responders in the periaqueductal gray (mean ± standard error; responders 43 ± 1 ms vs non-responders 32.5 ± 1 ms, p = 0.002) and the anterior cingulate cortex (mean ± standard error; responders 50 ± 1 ms vs non-responders 40 ± 1 ms, p = 0.01). Conclusions Erenumab response is associated with higher T2* in the periaqueductal gray and anterior cingulate cortex, regions that participate in pain processing and modulation. T2* differences between erenumab responders vs non-responders, a measure of brain iron accumulation, are seen at eight weeks post-treatment. Less iron accumulation in the periaqueductal gray and anterior cingulate cortex might play a role in the therapeutic mechanisms of migraine reduction associated with erenumab.

Abstract Objective To explore alterations in thalamic subfield volume and iron accumulation in individuals with post‐traumatic headache (PTH) relative to healthy controls. Background The thalamus plays a pivotal role in the pathomechanism of pain and headache, yet the role of the thalamus in PTH attributed to mild traumatic brain injury (mTBI) remains unclear. Methods A total of 107 participants underwent multimodal T1‐weighted and T2* brain magnetic resonance imaging. Using a clinic‐based observational study, thalamic subfield volume and thalamic iron accumulation were explored in 52 individuals with acute PTH (mean age = 41.3; standard deviation [SD] = 13.5), imaged on average 24 days post mTBI, and compared to 55 healthy controls (mean age = 38.3; SD = 11.7) without history of mTBI or migraine. Symptoms of mTBI and headache characteristics were assessed at baseline (0–59 days post mTBI) ( n = 52) and 3 months later ( n = 46) using the Symptom Evaluation of the Sports Concussion Assessment Tool (SCAT‐5) and a detailed headache history questionnaire. Results Relative to controls, individuals with acute PTH had significantly less volume in the lateral geniculate nucleus (LGN) (mean volume: PTH = 254.1, SD = 43.4 vs. controls = 278.2, SD = 39.8; p = 0.003) as well as more iron deposition in the left LGN (PTH: T2* signal = 38.6, SD = 6.5 vs. controls: T2* signal = 45.3, SD = 2.3; p = 0.048). Correlations in individuals with PTH revealed a positive relationship between left LGN T2* iron deposition and SCAT‐5 symptom severity score at baseline ( r = −0.29, p = 0.019) and maximum headache intensity at the 3‐month follow‐up ( r = −0.47, p = 0.002). Conclusion Relative to healthy controls, individuals with acute PTH had less volume and higher iron deposition in the left LGN. Higher iron deposition in the left LGN might reflect mTBI severity and poor headache recovery.

AbstractData-driven machine-learning methods on neuroimaging (e.g. MRI) are of great interest for the investigation and classification of neurological diseases. However, traditional machine learning requires domain knowledge to delineate the brain regions first, followed by feature extraction from the regions. Compared with this semi-automated approach, recently developed deep learning methods have advantages since they do not require such prior knowledge; instead, deep learning methods can automatically find features that differentiate MRIs from different cohorts. In the present study, we developed a deep learning-based classification pipeline distinguishing brain MRIs of individuals with one of three types of headaches [migraine (n = 95), acute post-traumatic headache (n = 48) and persistent post-traumatic headache (n = 49)] from those of healthy controls (n = 532) and identified the brain regions that most contributed to each classification task. Our pipeline included: (i) data preprocessing; (ii) binary classification of healthy controls versus headache type using a 3D ResNet-18; and (iii) biomarker extraction from the trained 3D ResNet-18. During the classification at the second step of our pipeline, we resolved two common issues in deep learning methods, limited training data and imbalanced samples from different categories, by incorporating a large public data set and resampling among the headache cohorts. Our method achieved the following classification accuracies when tested on independent test sets: (i) migraine versus healthy controls—75% accuracy, 66.7% sensitivity and 83.3% specificity; (2) acute post-traumatic headache versus healthy controls—75% accuracy, 66.7% sensitivity and 83.3% specificity; and (3) persistent post-traumatic headache versus healthy controls—91.7% accuracy, 100% sensitivity and 83.3% specificity. The most significant biomarkers identified by the classifier for migraine were caudate, caudal anterior cingulate, superior frontal, thalamus and ventral diencephalon. For acute post-traumatic headache, lateral occipital, cuneus, lingual, pericalcarine and superior parietal regions were identified as most significant biomarkers. Finally, for persistent post-traumatic headache, the most significant biomarkers were cerebellum, middle temporal, inferior temporal, inferior parietal and superior parietal. In conclusion, our study shows that the deep learning methods can automatically detect aberrations in the brain regions associated with different headache types. It does not require any human knowledge as input which significantly reduces human effort. It uncovers the great potential of deep learning methods for classification and automatic extraction of brain imaging–based biomarkers for these headache types.

e16550 Background: Improved computational power and modern algorithms have generated significant interest in radiomics for cancer diagnosis and staging. Here we assess the performance of deep learning (DL) models as a means for feature extraction in combination with supervised machine learning (ML) algorithms for accurate staging and chemotherapy response assessment of bladder cancer. Methods: Deidentified grayscale CT images from bladder cancer patients scheduled to undergo radical cystectomy were included in this retrospective study. These images were manually annotated with two regional masks (normal region and cancer region). Five DL models- namely, AlexNet, GoogleNet, InceptionV3, ResNet-50, and XceptionNet pre-trained on the ImageNet dataset, a public dataset, were then fine-tuned on our bladder CT scan data to extract features. Through feature selection process, the subset of the features was used to build ML classifiers for classification. The classification was performed using five different ML classifiers, namely k-Nearest Neighbor (KNN), Naïve-Bayes (NB), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), and Decision Tree (DT). The classification task was performed with 10-fold cross-validation, and each of the experiments contained a different but not mutually exclusive subset of samples. The evaluation metrics include accuracy, sensitivity, specificity, precision, and F1-score. Results: A total of 200 deidentified grayscale CT images of 100 patients with histologically proven bladder cancer, were included in this study. For experiment (1) normal vs. cancer, the LDA classifier on XceptionNet based features provides the best performance with an accuracy of 86.07%, sensitivity of 96.75%, specificity of 69.65%, precision of 83.07%, and F1-score of 89.39%. For experiment (2) non-muscle invasive Cancer (NMIBC) vs. muscle invasive bladder cancer (MIBC), the LDA classifier on XceptionNet based features provided the best performance with an accuracy of 79.72%, sensitivity of 66.62%, specificity of 87.39%, precision of 75.58%, and F1-score of 70.81%. For experiment (3) T0 lesion vs. MIBC, the LDA classifier on XceptionNet based features provides the best performance with an accuracy of 74.96%, sensitivity of 80.51%, specificity of 70.22%, precision of 69.78%, and F1-score of 74.73%. Conclusions: Our proposed model has shown good results in differentiating normal vs cancer and promising performance in differentiating T0 vs MIBC after chemotherapy treatment. We are expanding our dataset to further improve the performance in differentiating T0 vs MIBC. In addition, we will investigate the applicability of GAN for data augmentation to address data limit. We believe the hybrid DL and ML framework may facilitates radiologists' decisions and clinical decision-making in patients with bladder cancer.

AbstractIntroductionMultiple positron emission tomography (PET) tracers are available for amyloid imaging, posing a significant challenge to consensus interpretation and quantitative analysis. We accordingly developed and validated a deep learning model as a harmonization strategy.MethodA Residual Inception Encoder‐Decoder Neural Network was developed to harmonize images between amyloid PET image pairs made with Pittsburgh Compound‐B and florbetapir tracers. The model was trained using a dataset with 92 subjects with 10‐fold cross validation and its generalizability was further examined using an independent external dataset of 46 subjects.ResultsSignificantly stronger between‐tracer correlations (P < .001) were observed after harmonization for both global amyloid burden indices and voxel‐wise measurements in the training cohort and the external testing cohort.DiscussionWe proposed and validated a novel encoder‐decoder based deep model to harmonize amyloid PET imaging data from different tracers. Further investigation is ongoing to improve the model and apply to additional tracers.

Objectives Although iron accumulation in pain-processing brain regions has been associated with repeated migraine attacks, brain structural changes associated with post-traumatic headache have yet to be elucidated. To determine whether iron accumulation is associated with acute post-traumatic headache, magnetic resonance transverse relaxation rates (T2*) associated with iron accumulation were investigated between individuals with acute post-traumatic headache attributed to mild traumatic brain injury and healthy controls. Methods Twenty individuals with acute post-traumatic headache and 20 age-matched healthy controls underwent 3T brain magnetic resonance imaging including quantitative T2* maps. T2* differences between individuals with post-traumatic headache versus healthy controls were compared using age-matched paired t-tests. Associations of T2* values with headache frequency and number of mild traumatic brain injuries were investigated using multiple linear regression in individuals with post-traumatic headache. Significance was determined using uncorrected p-value and cluster size threshold. Results Individuals with post-traumatic headache had lower T2* values compared to healthy controls in cortical (bilateral frontal, bilateral anterior and posterior cingulate, right postcentral, bilateral temporal, right supramarginal, right rolandic, left insula, left occipital, right parahippocampal), subcortical (left putamen, bilateral hippocampal) and brainstem regions (pons). Within post-traumatic headache subjects, multiple linear regression showed a negative association between T2* in the right inferior parietal/supramarginal regions and number of mild traumatic brain injuries and a negative association between T2* in bilateral cingulate, bilateral precuneus, bilateral supplementary motor areas, bilateral insula, right middle temporal and right lingual areas and headache frequency. Conclusions Acute post-traumatic headache is associated with iron accumulation in multiple brain regions. Correlations with headache frequency and number of lifetime mild traumatic brain injuries suggest that iron accumulation is part of the pathophysiology or a marker of mild traumatic brain injury and post-traumatic headache.

Abstract Background/Objective: Changes in speech can be detected objectively before and during migraine attacks. The goal of this study was to interrogate whether speech changes can be detected in subjects with post-traumatic headache (PTH) attributed to mild traumatic brain injury (mTBI) and whether there are within-subject changes in speech during headaches compared to the headache-free state. Methods: Using a series of speech elicitation tasks uploaded via a mobile application, PTH subjects and healthy controls (HC) provided speech samples once every three days, over a period of 12 weeks. The following speech parameters were assessed: vowel space area, vowel articulation precision, consonant articulation precision, average pitch, pitch variance, speaking rate and pause rate. Speech samples of subjects with PTH were compared to HC. To assess speech changes associated with PTH, speech samples of subjects during headache were compared to speech samples when subjects were headache-free. All analyses were conducted using a mixed-effect model design.Results: Longitudinal speech samples were collected from nineteen subjects with PTH (mean age=42.5, SD=13.7) who were an average of 14 days (SD=32.2) from their mTBI at the time of enrollment and thirty-one HC (mean age=38.7, SD=12.5). Regardless of headache presence or absence, PTH subjects had longer pause rates and reductions in vowel and consonant articulation precision relative to HC. On days when speech was collected during a headache, there were longer pause rates, slower sentence speaking rates and less precise consonant articulation compared to the speech production of HC. During headache, PTH subjects had slower speaking rates and altered vowel articulation compared to when they were headache-free. Conclusions: Compared to HC, subjects with acute PTH demonstrate altered speech as measured by objective features of speech production. For individuals with PTH, speech was altered during headache vs. when they were headache-free, suggesting that speech alterations were related to PTH and not solely due to the underlying mTBI.

Abstract Prostate cancer is the second most common new cancer diagnosis in the United States. The prostate gland sits beneath the urinary bladder and surrounds the first part of the urethra. Usually, prostate cancer is slow-growing; stays confined to the prostate gland; and can be treated conservatively (active surveillance) or with surgery. However, if the cancer has spread beyond the prostate, such as to the lymph nodes, then that suggests the cancer is more aggressive and surgery is not adequate. In those cases, radiation and/or systemic therapies (e.g., chemotherapy, immunotherapy) are required. The challenge is that it is often difficult for radiologists to differentiate malignant lymph nodes from non-malignant ones with current medical imaging technology. In this study, we design a scalable hybrid approach utilizing a deep learning model to extract features into a machine learning classifier to automatically identify malignant lymph nodes in patients with prostate cancer.