China
Africa
Explore chapters and articles related to this topic
Tumor segmentation
Published in Ruijiang Li, Lei Xing, Sandy Napel, Daniel L. Rubin, Radiomics and Radiogenomics, 2019
Spyridon Bakas, Rhea Chitalia, Despina Kontos, Yong Fan, Christos Davatzikos
Non-invasive, longitudinal measures capturing the complete spatial tumor extent and peritumoral tissue would allow for more comprehensive, hence accurate, characterization of the heterogeneity of the tumor, which would subsequently influence personalized prognosis and treatment. Imaging is a non-invasive and widely available method for assessing tumor status in vivo macroscopically, and hence, has a very promising role toward these goals. In recent years, imaging features describing the texture (also known as radiomics) have been included in diagnostic imaging reporting and data systems (IRADS) for (i) breast (i.e., BI-RADS) [20], (ii) prostate (i.e., PI-RADS) [21], and (iii) lung (i.e., LI-RADS) [22]. Furthermore, there has been mounting evidence that computational analysis of quantitative imaging phenomic features (a group of which are often called radiomic features) extracted from multi-parametric radiographic imaging modalities can characterize tumors comprehensively and provide critical information about various biological processes in the tumor microenvironment, as well as yield associations with the underlying molecular characteristics of the cancer (often referred to as radiogenomics) [23–38]. Since cancer patients are routinely scanned radiographically (typically before surgery and at multiple time points during treatment and follow-up), the availability of radiomic and radiogenomic imaging biomarkers would help evaluate the spatial and temporal heterogeneity of tumors.
Biological Imaging and Radiobiological Modeling for Treatment Planning and Response Assessment in Radiation Therapy
Published in Siyong Kim, John Wong, Advanced and Emerging Technologies in Radiation Oncology Physics, 2018
Vitali Moiseenko, Stephen R. Bowen, John P. Kirkpatrick, Robert Jeraj, Lawrence B. Marks
Radiobiological descriptions of normal tissue and tumor response to radiation are often limited to fitting the mean or median of population-based data with little to no consideration of patient-specific variability of tumor or normal tissue properties that modulate this response. Biological imaging, including both molecular and functional imaging, noninvasively investigates properties that are spatially localized either to cancerous or functional tissue and that may dynamically vary with time. To use biological imaging for therapeutic applications, such as treatment planning and treatment response evaluation of spatial and temporal variations in biological properties that are unique to each individual patient, quantitative imaging biomarkers must be established. Quantitative imaging biomarkers may be associated with clinical endpoints, and they may complement tissues biomarkers, both potentially driving precision medicine (Mirnezami et al., 2012). Conceptually, predictive imaging biomarkers might guide initial therapy and subsequent adaptive therapies, thus facilitating administration of the right therapy to an individual patient at the optimal time. Within precision radiation therapy, quantitative biological imaging can define treatment-planning targets, define functional avoidance regions, and assess the response of targets and functional tissues during and after therapy (Jeraj et al., 2015).
New Applications of Super-Resolution in Medical Imaging
Published in Peyman Milanfar, Super-Resolution Imaging, 2017
M. Dirk Robinson, Stephanie J. Chiu, Cynthia A. Toth, Joseph A. Izatt, Joseph Y. Lo, Sina Farsiu
Early, fast, and accurate detection of imaging biomarkers of the onset and progression of diseases is of great importance to the medical community since early detection and intervention often results in optimal treatment and recovery. The advent of novel imaging systems has for the first time enabled clinicians and medical researchers to visualize the anatomical substructures, pathology, and functional features in vivo. However, earlier biomarkers of disease onset are often critically smaller or weaker in contrast compared to their corresponding features in the advanced stages of disease. Therefore, medical imaging community strives for inventing higher-resolution/contrast imaging systems. As noted in Section 13.2, super-resolution can be beneficial in improving the image quality of many medical imaging systems without the need for significant hardware alternation.
A novel solution of an elastic net regularisation for dementia knowledge discovery using deep learning
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2023
Kshitiz Shrestha, Omar Hisham Alsadoon, Abeer Alsadoon, Tarik A. Rashid, Rasha S. Ali, P.W.C. Prasad, Oday D. Jerew
(J. J. Liu et al., 2016) developed the artificial neural network that uses multi-view fusion that combines with multiple estimators. This research paper provides a two-stage binary classification, i.e., Alzheimer’s Disease and Normal Control and Mild Cognitive Impairment and Normal Control. Ensemble learning layer and the soft-max layer is present in it to predict either the given subject is Alzheimer’s Disease or Mild Cognitive Impairment. Among the three views Magnetic Resonance Imaging + Positron Emission Tomography gives the best result. However, the proposed solution only focuses on binary classification, so it is necessary to extend the framework to the multi-class classification problem. (W. J. Niessen, 2016) studied the strong influence of genetics in the brain for knowledge discovery from Magnetic Resonance Imaging images. The author uses the extracted data by data references, the ageing model, and computer-aided for the diagnosis. The proposed solution makes analyzes to the brain Magnetic Resonance Imaging images in various aspects of a brain using the methodology like Imaging genetics, Ageing brain models, quantitative imaging biomarker, and Machine learning. However, the proposed solution has not done any measurement which should be considered to get the accuracy for comparison purposes.
Early Prediction of Progression to Alzheimer’s Disease using Multi-Modality Neuroimages by a Novel Ordinal Learning Model ADPacer
Published in IISE Transactions on Healthcare Systems Engineering, 2023
Lujia Wang, Zhiyang Zheng, Yi Su, Kewei Chen, David Weidman, Teresa Wu, ShihChung Lo, Fleming Lure, Jing Li
The contributions of this paper are summarized as follows:Our study focused on predicting the pace of MCI progression/conversion to AD by integrating multi-modality neuroimaging and non-imaging datasets. This complements the existing studies in literature that mainly focus on binary classification of converters and non-converters. Prediction of progression pace has multifold benefits as it would allow for individually tailored intervention, better preparation of patients and caregivers, and more nuanced patient selection strategies in clinical trials.We proposed a novel ADPacer model to leverage training samples with label ambiguity to augment the training set with precisely-labeled samples. This capability differentiates ADPacer from existing ordinal learning algorithms.We applied ADPacer to MCI patient cohorts from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and Australian Imaging, Biomarker & Lifestyle Flagship Study of Ageing (AIBL), and demonstrated the superior performance of ADPacer compared to existing ordinal learning algorithms. We also integrated the SHapley Additive exPlanations (SHAP) method with ADPacer to assess the contributions from different modalities to the model prediction. The findings are consistent with the AD literature.
Machine learning-based lungs cancer detection using reconstruction independent component analysis and sparse filter features
Published in Waves in Random and Complex Media, 2021
Lal Hussain, Majid Saeed Almaraashi, Wajid Aziz, Nazneen Habib, Saif-Ur-Rehman Saif Abbasi
To determine the complexity of a system, a number of independent variables are required to predict or produce the output of the system. For example, in the skeletal muscle, the fiber type expression is dependent on the hormonal, genetic, neuronal, cardiovascular and activity-related influences that affect the fast or slow myosin isoform expression in the muscle [54,55]. There are several examples of reduction of structural components with aging disease including collagen fibers per unit of surface area in skin tissue [56], estrogen hormone activity in females [57], sinus node cells in the heart [58], the number of alveoli of the lungs [59,60], physical brain damage or even death may happen due to the frequent occurrence of seizures [61,62], chronic alcohol drinking cause alcohol-related brain damage including brain structure change [63]. In this study, we used approximate entropy, sample entropy along with its variants using K-Dimensional (KD) tree approach and wavelet entropic measures to distinguish the lung cancer type SCLC from NSCLC. The entropy-based results reveal promising detection outcome to predict the lung cancer, which is an indication that entropy is an encouraging quantitative imaging biomarker to characterize cancer imaging phenotype.