Radiomics-Clinical AI Model with Probability Weighted Strategy for Prognosis Prediction in Non-Small Cell Lung Cancer.

In this study, we propose a radiomics clinical probability-weighted model for the prediction of prognosis for non-small cell lung cancer (NSCLC). The model combines radiomics features extracted from radiotherapy (RT) planning images with clinical factors such as age, gender, histology, and tumor stage. CT images with radiotherapy structures of 422 NSCLC patients were retrieved from The Cancer Imaging Archive (TCIA). Radiomic features were extracted from gross tumor volumes (GTVs). Five machine learning algorithms, namely decision trees (DT), random forests (RF), extreme boost (EB), support vector machine (SVM) and generalized linear model (GLM) were optimized by a voted ensemble machine learning (VEML) model. A probabilistic weighted approach is used to incorporate the uncertainty associated with both radiomic and clinical features and to generate a probabilistic risk score for each patient. The performance of the model is evaluated using a receiver operating characteristic (ROC). The Radiomic model, clinical factor model, and combined radiomic clinical probability-weighted model demonstrated good performance in predicting NSCLC survival with AUC of 0.941, 0.856 and 0.949, respectively. The combined radiomics clinical probability-weighted enhanced model achieved significantly better performance than the radiomic model in 1-year survival prediction (chi-square test, p < 0.05). The proposed model has the potential to improve NSCLC prognosis and facilitate personalized treatment decisions.

Prognostic Prediction of Cancer Based on Radiomics Features of Diagnostic Imaging: The Performance of Machine Learning Strategies.

Tumor phenotypes can be characterized by radiomics features extracted from images. However, the prediction accuracy is challenged by difficulties such as small sample size and data imbalance. The purpose of the study was to evaluate the performance of machine learning strategies for the prediction of cancer prognosis. A total of 422 patients diagnosed with non-small cell lung carcinoma (NSCLC) were selected from The Cancer Imaging Archive (TCIA). The gross tumor volume (GTV) of each case was delineated from the respective CT images for radiomic features extraction. The samples were divided into 4 groups with survival endpoints of 1 year, 3 years, 5 years, and 7 years. The radiomic image features were analyzed with 6 different machine learning methods: decision tree (DT), boosted tree (BT), random forests (RF), support vector machine (SVM), generalized linear model (GLM), and deep learning artificial neural networks (DL-ANNs) with 70:30 cross-validation. The overall average prediction performance of the BT, RF, DT, SVM, GLM and DL-ANNs was AUC with 0.912, 0.938, 0.793, 0.746, 0.789 and 0.705 respectively. The RF and BT gave the best and second performance in the prediction. The DL-ANN did not show obvious advantage in predicting prognostic outcomes. Deep learning artificial neural networks did not show a significant improvement than traditional machine learning methods such as random forest and boosted trees. On the whole, the accurate outcome prediction using radiomics serves as a supportive reference for formulating treatment strategy for cancer patients.

Radiomics from Various Tumour Volume Sizes for Prognosis Prediction of Head and Neck Squamous Cell Carcinoma: A Voted Ensemble Machine Learning Approach.

Background: Traditionally, cancer prognosis was determined by tumours size, lymph node spread and presence of metastasis (TNM staging). Radiomics of tumour volume has recently been used for prognosis prediction. In the present study, we evaluated the effect of various sizes of tumour volume. A voted ensemble approach with a combination of multiple machine learning algorithms is proposed for prognosis prediction for head and neck squamous cell carcinoma (HNSCC). Methods: A total of 215 HNSCC CT image sets with radiotherapy structure sets were acquired from The Cancer Imaging Archive (TCIA). Six tumour volumes, including gross tumour volume (GTV), diminished GTV, extended GTV, planning target volume (PTV), diminished PTV and extended PTV were delineated. The extracted radiomics features were analysed by decision tree, random forest, extreme boost, support vector machine and generalized linear algorithms. A voted ensemble machine learning (VEML) model that optimizes the above algorithms was used. The receiver operating characteristic area under the curve (ROC-AUC) were used to compare the performance of machine learning methods, including accuracy, sensitivity and specificity. Results: The VEML model demonstrated good prognosis prediction ability for all sizes of tumour volumes with reference to GTV and PTV with high accuracy of up to 88.3%, sensitivity of up to 79.9% and specificity of up to 96.6%. There was no significant difference between the various target volumes for the prognostic prediction of HNSCC patients (chi-square test, p > 0.05). Conclusions: Our study demonstrates that the proposed VEML model can accurately predict the prognosis of HNSCC patients using radiomics features from various tumour volumes.

Radiomics-Based Artificial Intelligence Differentiation of Neurodegenerative Diseases with Reference to the Volumetry.

This study aimed to build automated detection models-one by brain regional volume (V-model), and the other by radiomics features of the whole brain (R-model)-to differentiate mild cognitive impairment (MCI) from cognitive normal (CN), and Alzheimer's Disease (AD) from mild cognitive impairment (MCI). The objectives are to compare the models and identify whether radiomics or volumetry can provide a better prediction for differentiating different types of dementia. METHOD: 582 MRI T1-weighted images were retrieved from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, which is a multicenter operating open source database for AD. In total, 97 images of AD, 293 images of MCI patient and 192 images of cognitive normal were divided into a training, a validation and a test group at a ratio of 70:15:15. For each T1-weighted image, volumetric segmentation was performed with the image analysis software FreeSurfer, and radiomics features were retrieved by imaging research software 3D slicers. Brain regional volume and radiomics features were used to build the V-model and R-model, respectively, using the random forest algorithm by R. The receiver operating characteristics (ROC) curve of both models were used to evaluate their diagnostic accuracy and reliability to differentiate AD, MCI and CN. RESULTS: To differentiate MCI and CN, both V-model and R-model achieved excellent performance, with an AUC of 0.9992 ± 0.0022 and 0.9850 ± 0.0032, respectively. No significant difference was found between the two AUCs, indicating both models attained similar good performance. In MCI and AD differentiation, the V-model and R-model yielded AUC of 0.9986 ± 0.0013 and 0.9714 ± 0.0175, respectively. The best performance was to differentiate AD from CN, where the V-model and R-model yielded AUC of 0.9994 ± 0.0019 and 0.9830 ± 0.009, respectively. The results suggested that both volumetry and radiomics approaches could be used in differentiating AD, MCI and CN, based on T1 weighted MR images using random forest algorithm successfully. CONCLUSION: This study showed that the radiomics features from T1-weighted MR images achieved excellence performance in differentiating AD, MCI and CN. Compared to the volumetry method, the accuracy, sensitivity and specificity are slightly lower in using radiomics, but still attained very good and reliable classification of the three stages of neurodegenerations. In view of the convenience and operator independence in feature extraction, radiomics can be a quantitative biomarker to differentiate the disease groups.

Multimodal Patient-Specific Registration for Breast Imaging Using Biomechanical Modeling with Reference to AI Evaluation of Breast Tumor Change.

BACKGROUND: The strategy to combat the problem associated with large deformations in the breast due to the difference in the medical imaging of patient posture plays a vital role in multimodal medical image registration with artificial intelligence (AI) initiatives. How to build a breast biomechanical model simulating the large-scale deformation of soft tissue remains a challenge but is highly desirable. METHODS: This study proposed a hybrid individual-specific registration model of the breast combining finite element analysis, property optimization, and affine transformation to register breast images. During the registration process, the mechanical properties of the breast tissues were individually assigned using an optimization process, which allowed the model to become patient specific. Evaluation and results: The proposed method has been extensively tested on two datasets collected from two independent institutions, one from America and another from Hong Kong. CONCLUSIONS: Our method can accurately predict the deformation of breasts from the supine to prone position for both the Hong Kong and American samples, with a small target registration error of lesions.

Accurate Age Determination for Adolescents Using Magnetic Resonance Imaging of the Hand and Wrist with an Artificial Neural Network-Based Approach.

This study proposes an accurate method in assessing chronological age of the adolescents using a machine learning approach using MRI images. We also examined the value of MRI with Tanner-Whitehouse 3 (TW3) method in assessing skeletal maturity. Seventy-nine 12-17-year-old healthy Hong Kong Chinese adolescents were recruited. The left hand and wrist region were scanned by a dedicated skeletal MRI scanner. T1-weighted three-dimensional coronal view images for the left hand and wrist region were acquired. Independent maturity indicators such as subject body height, body weight, bone marrow composition intensity quantified by MRI, and TW3 skeletal age were included for artificial neural network (ANN) analysis. Our results indicated that the skeletal age was generally underestimated using TW3 method, and significant difference (p < 0.05) was noted for skeletal age with chronological age for female category and at later stage of adolescence (15 to 17 years old) in both genders. In our proposed machine learning approach, ages determined by ANN method agreed well with chronological age (p > 0.05).The machine learning approach using ANN method was about 10-fold more accurate than the TW3 method using MRI alone. It offers a more objective and accurate solution for prospective chronological maturity assessment for adolescents.

Development of a prototype of the tele-localisation system in radiotherapy using personal digital assistant via wireless communication.

In localisation of radiotherapy treatment field, the oncologist is present at the simulator to approve treatment details produced by the therapist. Problems may arise if the oncologist is not available and the patient requires urgent treatment. The development of a tele-localisation system is a potential solution, where the oncologist uses a personal digital assistant (PDA) to localise the treatment field on the image sent from the simulator through wireless communication and returns the information to the therapist after his or her approval. Our team developed the first tele-localisation prototype, which consisted of a server workstation (simulator) for the administration of digital imaging and communication in medicine localisation images including viewing and communication with the PDA via a Wi-Fi network; a PDA (oncologist's site) installed with the custom-built programme that synchronises with the server workstation and performs treatment field editing. Trial tests on accuracy and speed of the prototype system were conducted on 30 subjects with the treatment regions covering the neck, skull, chest and pelvis. The average time required in performing the localisation using the PDA was less than 1.5 min, with the blocked field longer than the open field. The transmission speed of the four treatment regions was similar. The average physical distortion of the images was within 4.4% and the accuracy of field size indication was within 5.3%. Compared with the manual method, the tele-localisation system presented with an average deviation of 5.5%. The prototype system fulfilled the planned objectives of tele-localisation procedure with reasonable speed and accuracy.

Effects of Panax ginseng-containing herbal plasters on compressed intervertebral discs in an in vivo rat tail model.

BACKGROUND: Tienchi (Panax notoginseng) has been used in conservative treatments for back pain as a major ingredient of many herbal medicines. This study aims to investigate the effects of a herbal medicine containing tienchi on compressed intervertebral discs in rats. METHODS: Using an in vivo rat tail model, intervertebral disc compression was simulated in the caudal 8-9 discs of 25 rats by continuous static compression (11 N) for 2 weeks. An herbal medicine plaster (in which the major ingredient was tienchi) was externally applied to the compressed disc (n=9) for three weeks, and held in place by an adhesive bandage, in animals in the Chinese Medicine (CM) group. The effect of the bandage was evaluated in a separate placebo group (n=9), while no intervention with unrestricted motion was provided to rats in an additional control group (n=7). Disc structural properties were quantified by in vivo disc height measurement and in vitro morphological analysis. RESULTS: Disc height decreased after the application of compression (P < 0.001). The disc height decreased continuously in the control (P = 0.006) and placebo (P = 0.003) groups, but was maintained in the CM group (P = 0.494). No obvious differences in disc morphology were observed among the three groups (P = 0.896). CONCLUSION: The tienchi-containing herbal plaster had no significant effect on the morphology of compressed discs, but maintained disc height in rats.

An image feature approach for computer-aided detection of ischemic stroke.

We present a computer-aided detection (CAD) scheme for early detection of ischemic stroke with small lesions using image feature characteristics. A novel Circular Adaptive Region of Interest (CAROI) method is proposed to analyze the Computed Tomography (CT) images of the brain. Our result indicates that for the emergency physicians and radiology residents, there is a significant improvement in sensitivity and specificity when using CAD (P < 0.005). A mathematical model is established incorporating the weighting of the feature changes. Our CAD scheme is promising for early detection of ischemic stroke and helps improve the efficiency and accuracy of clinical practice.

Computer-generated index for evaluation of idiopathic scoliosis in digital chest images: a comparison with digital measurement.

An intelligence system was used to generate index for scoliosis. Tests were designed to evaluate the consistency of the automatic computer-generated index and to quantify the correlation between Cobb angle and computer generated scoliosis classification index (SCI). A fully automatic computer-generated index can be used to assess the extent of spinal curvature rather than manual measurement on radiographs. This study aims to evaluate the relation of an automatic computer-generated index in assessing the spinal curvature of scoliosis quantitatively on digital chest images. Sixty chest radiographs were obtained in this study. Cobb angle measurement and the index generated were compared by parametric statistical tests. The SCI method was demonstrated to be reproducible. There was also statically significant positive correlation between Cobb angle and SCI (Pearson's correlation: r = 0.9229). The Computer-generated index method is valid and reliable in quantifying measurement of spinal curvature of scoliosis as the correlation between Cobb's angle and SCI in nearly perfect positive for Cobb angle more than 10 degree. It is noted that with widely use of this computer method, this quantitative method proposed is a promising method in improving the reliability of scoliosis assessment and reducing the workload of clinical staff.

The calculation of dose enhancement close to platinum implants for skull radiography.

Materials with high atomic numbers experience the occurrence of the photoelectric effect when they are irradiated by low energy photons. A short range dose enhancement, due to the dominant photoelectric effect, close to platinum implants (Z = 78) in diagnostic radiography cannot be easily measured experimentally. The enhanced dose may increase the risk for adverse health effects from cancer or may damage vital brain structures close to the high atomic number implants. In the present work, Monte Carlo simulation using the LSCAT version of PRESTA EGS4 was employed to investigate the resulting dose enhancements. The results show that the highest estimated dose enhancement of 79% for brain tissues close to platinum implants was calculated for 65 kV x-ray energy and 180% for 120 kV x-ray energy.

A mobile phone integrated health care delivery system of medical images.

With the growing computing capability of mobile phones, a handy mobile controller is developed for accessing the picture archiving and communication system (PACS) to enhance image management for clinicians with nearly no restriction in time and location using various wireless communication modes. The PACS is an integrated system for the distribution and archival of medical images that are acquired by different imaging modalities such as CT (computed tomography) scanners, CR (computed radiography) units, DR (digital radiography) units, US (ultrasonography) scanners, and MR (magnetic resonance) scanners. The mobile controller allows image management of the PACS including display, worklisting, query and retrieval of medical images in DICOM format. In this mobile system, a server program is developed in a PACS Web server which serves as an interface for client programs in the mobile phone and the enterprise PACS for image distribution in hospitals. The application processing is performed on the server side to reduce computational loading in the mobile device. The communication method of mobile phones can be adapted to multiple wireless environments in Hong Kong. This allows greater feasibility to accommodate the rapidly changing communication technology. No complicated computer hardware or software is necessary. Using a mobile phone embedded with the mobile controller client program, this system would serve as a tool for heath care and medical professionals to improve the efficiency of the health care services by speedy delivery of image information. This is particularly important in case of urgent consultation, and it allows health care workers better use of the time for patient care.