Byung-Chul Kim

This study aimed to identify a distant-recurrence image biomarker in NSCLC by investigating correlations between heterogeneity functional gene expression and fluorine-18-2-fluoro-2-deoxy-D-glucose positron emission tomography (¹⁸F-FDG PET) image features of NSCLC patients. RNA-sequencing data and ¹⁸F-FDG PET images of 53 patients with NSCLC (19 with distant recurrence and 34 without recurrence) from The Cancer Imaging Archive and The Cancer Genome Atlas Program databases were used in a combined analysis. Weighted correlation network analysis was performed to identify gene groups related to distant recurrence. Genes were selected for functions related to distant recurrence. In total, 47 image features were extracted from PET images as radiomics. The relationship between gene expression and image features was estimated using a hypergeometric distribution test with the Pearson correlation method. The distant recurrence prediction model was validated by a random forest (RF) algorithm using image texture features and related gene expression. In total, 37 gene modules were identified by gene-expression pattern with weighted gene co-expression network analysis. The gene modules with the highest significance were selected (p-value < 0.05). Nine genes with high protein–protein interaction and area under the curve (AUC) were identified as hub genes involved in the proliferation function, which plays an important role in distant recurrence of cancer. Four image features (GLRLM_SRHGE, GLRLM_HGRE, SUVmean, and GLZLM_GLNU) and six genes were identified to be correlated (p-value < 0.1). AUCs (accuracy: 0.59, AUC: 0.729) from the 47 image texture features and AUCs (accuracy: 0.767, AUC: 0.808) from hub genes were calculated using the RF algorithm. AUCs (accuracy: 0.783, AUC: 0.912) from the four image texture features and six correlated genes and AUCs (accuracy: 0.738, AUC: 0.779) from only the four image texture features were calculated using the RF algorithm. The four image texture features validated by heterogeneity group gene expression were found to be related to cancer heterogeneity. The identification of these image texture features demonstrated that advanced prediction of NSCLC distant recurrence is possible using the image biomarker.

Motion estimation and compensation are necessary for improvement of tumor quantification analysis in positron emission tomography (PET) images. The aim of this study was to propose adaptive PET imaging with internal motion estimation and correction using regional artificial evaluation of tumors injected with low-dose and high-dose radiopharmaceuticals. In order to assess internal motion, molecular sieves imitating tumors were loaded with ¹⁸F and inserted into the lung and liver regions in rats. All models were classified into two groups, based on the injected radiopharmaceutical activity, to compare the effect of tumor intensity. The PET study was performed with injection of F-18 fluorodeoxyglucose (¹⁸F-FDG). Respiratory gating was carried out by external trigger device. Count, signal to noise ratio (SNR), contrast and full width at half maximum (FWHM) were measured in artificial tumors in gated images. Motion correction was executed by affine transformation with estimated internal motion data. Monitoring data were different from estimated motion. Contrast in the low-activity group was 3.57, 4.08 and 6.19, while in the high-activity group it was 10.01, 8.36 and 6.97 for static, 4 bin and 8 bin images, respectively. The results of the lung target in 4 bin and the liver target in 8 bin showed improvement in FWHM and contrast with sufficient SNR. After motion correction, FWHM was improved in both regions (lung: 24.56%, liver: 10.77%). Moreover, with the low dose of radiopharmaceuticals the PET image visualized specific accumulated radiopharmaceutical areas in the liver. Therefore, low activity in PET images should undergo motion correction before quantification analysis using PET data. We could improve quantitative tumor evaluation by considering organ region and tumor intensity.

We compared the accuracy of prediction of the response to neoadjuvant chemotherapy (NAC) in osteosarcoma patients between machine learning approaches of whole tumor utilizing fluorine−¹⁸fluorodeoxyglucose (¹⁸F-FDG) uptake heterogeneity features and a convolutional neural network of the intratumor image region. In 105 patients with osteosarcoma, ¹⁸F-FDG positron emission tomography/computed tomography (PET/CT) images were acquired before (baseline PET0) and after NAC (PET1). Patients were divided into responders and non-responders about neoadjuvant chemotherapy. Quantitative ¹⁸F-FDG heterogeneity features were calculated using LIFEX version 4.0. Receiver operating characteristic (ROC) curve analysis of ¹⁸F-FDG uptake heterogeneity features was used to predict the response to NAC. Machine learning algorithms and 2-dimensional convolutional neural network (2D CNN) deep learning networks were estimated for predicting NAC response with the baseline PET0 images of the 105 patients. ML was performed using the entire tumor image. The accuracy of the 2D CNN prediction model was evaluated using total tumor slices, the center 20 slices, the center 10 slices, and center slice. A total number of 80 patients was used for k-fold validation by five groups with 16 patients. The CNN network test accuracy estimation was performed using 25 patients. The areas under the ROC curves (AUCs) for baseline PET maximum standardized uptake value (SUVmax), total lesion glycolysis (TLG), metabolic tumor volume (MTV), and gray level size zone matrix (GLSZM) were 0.532, 0.507, 0.510, and 0.626, respectively. The texture features test accuracy of machine learning by random forest and support vector machine were 0.55 and 0. 54, respectively. The k-fold validation accuracy and validation accuracy were 0.968 ± 0.01 and 0.610 ± 0.04, respectively. The test accuracy of total tumor slices, the center 20 slices, center 10 slices, and center slices were 0.625, 0.616, 0.628, and 0.760, respectively. The prediction model for NAC response with baseline PET0 texture features machine learning estimated a poor outcome, but the 2D CNN network using ¹⁸F-FDG baseline PET0 images could predict the treatment response before prior chemotherapy in osteosarcoma. Additionally, using the 2D CNN prediction model using a tumor center slice of ¹⁸F-FDG PET images before NAC can help decide whether to perform NAC to treat osteosarcoma patients.

Chemotherapy response and metastasis prediction play important roles in the treatment of pediatric osteosarcoma, which is prone to metastasis and has a high mortality rate. This study aimed to estimate the prediction model using gene expression and image texture features. 18F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) images of 52 pediatric osteosarcoma patients were used to estimate the machine learning algorithm. An appropriate algorithm was selected by estimating the machine learning accuracy. ¹⁸F-FDG PET/CT images of 21 patients were selected for prediction model development based on simultaneous KI67 and EZRIN expression. The prediction model for chemotherapy response and metastasis was estimated using area under the curve (AUC) maximum image texture features (AUC_max) and gene expression. The machine learning algorithm with the highest test accuracy in chemotherapy response and metastasis was selected using the random forest algorithm. The chemotherapy response and metastasis test accuracy with image texture features was 0.83 and 0.76, respectively. The highest test accuracy and AUC of chemotherapy response with AUC_max, KI67, and EZRIN were estimated to be 0.85 and 0.89, respectively. The highest test accuracy and AUC of metastasis with AUC_max, KI67, and EZRIN were estimated to be 0.85 and 0.8, respectively. The metastasis prediction accuracy increased by 10% using radiogenomics data.

Breast cancer metastasis can have a fatal outcome, with the prediction of metastasis being critical for establishing effective treatment strategies. RNA-sequencing (RNA-seq) is a good tool for identifying genes that promote and support metastasis development. The hub gene analysis method is a bioinformatics method that can effectively analyze RNA sequencing results. This can be used to specify the set of genes most relevant to the function of the cell involved in metastasis. Herein, a new machine learning model based on RNA-seq data using the random forest algorithm and hub genes to estimate the accuracy of breast cancer metastasis prediction. Single-cell breast cancer samples (56 metastatic and 38 non-metastatic samples) were obtained from the Gene Expression Omnibus database, and the Weighted Gene Correlation Network Analysis package was used for the selection of gene modules and hub genes (function in mitochondrial metabolism). A machine learning prediction model using the hub gene set was devised and its accuracy was evaluated. A prediction model comprising 54-functional-gene modules and the hub gene set (NDUFA9, NDUFB5, and NDUFB3) showed an accuracy of 0.769 ± 0.02, 0.782 ± 0.012, and 0.945 ± 0.016, respectively. The test accuracy of the hub gene set was over 93% and that of the prediction model with random forest and hub genes was over 91%. A breast cancer metastasis dataset from The Cancer Genome Atlas was used for external validation, showing an accuracy of over 91%. The hub gene assay can be used to predict breast cancer metastasis by machine learning.