A5063110838- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Advanced X-ray and CT Imaging
- Parkinson's Disease Mechanisms and Treatments
- Lung Cancer Diagnosis and Treatment
- MRI in cancer diagnosis
- Voice and Speech Disorders
- Head and Neck Cancer Studies
- Artificial Intelligence in Healthcare and Education
- Advanced Neuroimaging Techniques and Applications
- Neurological disorders and treatments
- Scientific and Engineering Research Topics
- Acute Ischemic Stroke Management
- Imbalanced Data Classification Techniques
- Sarcoma Diagnosis and Treatment
- Inflammatory Biomarkers in Disease Prognosis
- Lung Cancer Treatments and Mutations
- Colorectal and Anal Carcinomas
- Ferroptosis and cancer prognosis
- Gastric Cancer Management and Outcomes
- Brain Tumor Detection and Classification
- VLSI and FPGA Design Techniques
- Medical Imaging Techniques and Applications
- Explainable Artificial Intelligence (XAI)
- Pancreatic and Hepatic Oncology Research
University of British Columbia
2020-2025
University Hospital Carl Gustav Carus
2024
Helmholtz-Zentrum Dresden-Rossendorf
2024
Cardiff University
2024
TU Dresden
2024
National Center for Tumor Diseases
2024
University of Pennsylvania
2024
German Cancer Research Center
2024
Southern Medical University
2023
BC Cancer Agency
2022
We aimed to predict Montreal Cognitive Assessment (MoCA) scores in Parkinson's disease patients at year 4 using handcrafted radiomics (RF), deep (DF), and clinical (CF) features 0 (baseline) applied hybrid machine learning systems (HMLSs).
Although handcrafted radiomics features (RF) are commonly extracted via software, employing deep (DF) from learning (DL) algorithms merits significant investigation. Moreover, a "tensor'' paradigm where various flavours of given feature generated and explored can provide added value. We aimed to employ conventional tensor DFs, compare their outcome prediction performance RFs.
This study explores a semi-supervised learning (SSL), pseudo-labeled strategy using diverse datasets such as head and neck cancer (HNCa) to enhance lung (LCa) survival outcome predictions, analyzing handcrafted deep radiomic features (HRF/DRF) from PET/CT scans with hybrid machine systems (HMLSs). We collected 199 LCa patients both PET CT images, obtained TCIA our local database, alongside 408 HNCa images TCIA. extracted 215 HRFs 1024 DRFs by PySERA 3D autoencoder, respectively, within the...
Purpose: This study examines the core traits of image-to-image translation (I2I) networks, focusing on their effectiveness and adaptability in everyday clinical settings. Methods: We have analyzed data from 794 patients diagnosed with prostate cancer (PCa), using ten prominent 2D/3D I2I networks to convert ultrasound (US) images into MRI scans. also introduced a new analysis Radiomic features (RF) via Spearman correlation coefficient explore whether high performance (SSIM>85%) could detect...
Purpose: Lung cancer poses a significant global health challenge, necessitating improved prognostic methods for personalized treatment. This study introduces censor-aware semi-supervised learning (SSL) that integrates clinical and imaging data, addressing biases in traditional models handling censored data. Methods: We analyzed clinical, PET, CT data from 199 lung patients TCIA BC Cancer Agency, focusing on overall survival (OS) time as the primary outcome. Handcrafted (HRF) Deep Radiomics...
Medical imaging data frequently encounter image-generation heterogeneity and class imbalance properties, challenging strong generalized predictive performances with data-driven machine-learning methods. The purpose of this study was to investigate the impact harmonization oversampling methods on multi-center imbalanced datasets, specific application PET-based radiomics modeling for histologic subtype prediction in non-small cell lung cancer (NSCLC). retrospective included 245 patients...
We employed machine learning approaches to (I) determine distinct progression trajectories in Parkinson's disease (PD) (unsupervised clustering task), and (II) predict (supervised prediction from early (years 0 1) data, making use of clinical imaging features.We studied PD-subjects derived longitudinal datasets 0, 1, 2 & 4; Progressive Marker Initiative). extracted analyzed 981 features, including motor, non-motor, radiomics features for each region-of-interest (ROIs: left/right caudate...
In Parkinson's disease (PD), 5-10% of cases are genetic origin with mutations identified in several genes such as leucine-rich repeat kinase 2 (LRRK2) and glucocerebrosidase (GBA). We aim to predict these two gene using hybrid machine learning systems (HMLS), via imaging non-imaging data, the long-term goal conversion active disease.We studied 264 129 patients known LRRK2 GBA status from PPMI database. Each dataset includes 513 features clinical (CFs), conventional (CIFs) radiomic (RFs)...
There are currently no established disease modifying therapies for PD, and prediction of outcome in PD to power clinical studies is a very important area research. Assessment informed by imaging the dopamine system with transporter (DAT) single-photon emission computed tomography (SPECT) presence key symptoms. Recently, deep-learning based methods have shown promise medical image analysis tasks detection. The purpose this study was develop approach predict patients using longitudinal data...
Abstract Purpose To evaluate the impact of harmonization and oversampling methods on multi-center imbalanced datasets, with specific application to PET-based radiomics modeling for histologic subtype prediction in non-small cell lung cancer (NSCLC). Methods The study included 245 patients adenocarcinoma (ADC) 78 squamous carcinoma (SCC) from 4 centers. Utilizing 1502 features per patient, we trained, validated, externally tested machine-learning classifiers, investigate effect no (NoH) or...
Radiomics features hold significant value as quantitative imaging biomarkers for diagnosis, prognosis, and treatment response assessment. To generate radiomics ultimately develop signatures, various factors can be manipulated, including image discretization parameters (e.g., bin number or size), convolutional filters, segmentation perturbation, multi-modality fusion levels. Typically, only one set of is employed, resulting in a single "flavour" each feature. In contrast, we propose "tensor...
Radiomics features extract quantitative information from medical images, towards the derivation of biomarkers for clinical tasks, such as diagnosis, prognosis, or treatment response assessment. Different image discretization parameters (e.g. bin number size), convolutional filters, segmentation perturbation, multi-modality fusion levels can be used to generate radiomics and ultimately signatures. Commonly, only one set is used; resulting in value ‘flavour’ a given feature. We propose ‘tensor...
In the present work, we systematically probe a range of predictor machines (11 machines), and aim to find best combinations features result in improvements prediction outcome PD. First, created 32 18 conventional experimentally selected 4 arrangements 204 PD subjects. The were applied various thereby absolute error combination reached 4.3 (in MDS-UPDRS-III motor performance year 4). This is comparison previous works that attained errors around 9. second part, subset selector used for...