18F-FMISO PET-guided dose escalation with multifield optimization intensity-modulated proton therapy in nasopharyngeal carcinoma.

PURPOSE: The purpose of this study was to evaluate the radiotherapy planning feasibility of dose escalation with intensity-modulated proton therapy (IMPT) to hypoxic tumor regions identified on 18F-Fluoromisonidazole (FMISO) positron emission tomography and computed tomography (PET-CT) in NPC. MATERIALS AND METHODS: Nine patients with stages T3-4N0-3M0 NPC underwent 18F-FMISO PET-CT before and during week 3 of radiotherapy. The hypoxic volume (GTVhypo) is automatically generated by applying a subthresholding algorithm within the gross tumor volume (GTV) with a tumor to muscle standardized uptake value (SUV) ratio of 1.3 on the 18F-FMISO PET-CT scan. Two proton plans were generated for each patient, a standard plan to 70 Gy and dose escalation plan with upfront boost followed by standard 70GyE plan. The stereotactic boost was planned with single-field uniform dose optimization using two fields to deliver 10 GyE in two fractions to GTVhypo. The standard plan was generated with IMPT with robust optimization to deliver 70GyE, 60GyE in 33 fractions using simultaneous integrated boost technique. A plan sum was generated for assessment. RESULTS: Eight of nine patients showed tumor hypoxia on the baseline 18F-FMISO PET-CT scan. The mean hypoxic tumor volume was 3.9 cm3 (range .9-11.9cm3 ). The average SUVmax of the hypoxic volume was 2.2 (range 1.44-2.98). All the dose-volume parameters met the planning objectives for target coverage. Dose escalation was not feasible in three of eight patients as the D0.03cc of temporal lobe was greater than 75GyE. CONCLUSIONS: The utility of boost to the hypoxic volume before standard course of radiotherapy with IMPT is dosimetrically feasible in selected patients. Clinical trials are warranted to determine the clinical outcomes of this approach.

Multi-Instance Multi-Label Learning for Gene Mutation Prediction in Hepatocellular Carcinoma.

Gene mutation prediction in hepatocellular carcinoma (HCC) is of great diagnostic and prognostic value for personalized treatments and precision medicine. In this paper, we tackle this problem with multi-instance multi-label learning to address the difficulties on label correlations, label representations, etc. Furthermore, an effective oversampling strategy is applied for data imbalance. Experimental results have shown the superiority of the proposed approach.

Multi-Phase Cross-modal Learning for Noninvasive Gene Mutation Prediction in Hepatocellular Carcinoma.

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and the fourth most common cause of cancer-related death worldwide. Understanding the underlying gene mutations in HCC provides great prognostic value for treatment planning and targeted therapy. Radiogenomics has revealed an association between non-invasive imaging features and molecular genomics. However, imaging feature identification is laborious and error-prone. In this paper, we propose an end-to-end deep learning framework for mutation prediction in APOB, COL11A1 and ATRX genes using multiphasic CT scans. Considering intra-tumour heterogeneity (ITH) in HCC, multi-region sampling technology is implemented to generate the dataset for experiments. Experimental results demonstrate the effectiveness of the proposed model.

Liver tumor volume estimation by semi-automatic segmentation method.

Liver cancer is one of the most popular cancer diseases and causes a large amount of death every year. In order to make decisions such as liver resections, doctors will need to know the tumor volume, and further, the functional liver volume. Thus, an important task in radiology is the determination of tumor volume. Accurate segmentation of liver tumor from an abdominal image is one of the most important steps in 3D representation for liver volume measurement, liver transplant, and treatment planning[1]. Since manual segmenation is inconvenient, time consuming and depends on the individual operator to a large extent, automatic segmentation is much more preferred. In this paper, an active contour model is used to segment tumors from CT abdominal images. Initial boundary is manually placed by operators outside the tumor region. The snake deforms to the tumor boundary with the minimization of energy function. We then calculate the tumor volume using the series of segmented tumor slices. Results show that this method is quite efficient in tumor volume estimation compared with the WHO criteria, which measures the tumor by multiplying the longest perpendicular diameters.