Deep Learning Combined with Radiologist's Intervention Achieves Accurate Segmentation of Hepatocellular Carcinoma in Dual-Phase Magnetic Resonance Images.

Purpose: Segmentation of hepatocellular carcinoma (HCC) is crucial; however, manual segmentation is subjective and time-consuming. Accurate and automatic lesion contouring for HCC is desirable in clinical practice. In response to this need, our study introduced a segmentation approach for HCC combining deep convolutional neural networks (DCNNs) and radiologist intervention in magnetic resonance imaging (MRI). We sought to design a segmentation method with a deep learning method that automatically segments using manual location information for moderately experienced radiologists. In addition, we verified the viability of this method to assist radiologists in accurate and fast lesion segmentation. Method: In our study, we developed a semiautomatic approach for segmenting HCC using DCNN in conjunction with radiologist intervention in dual-phase gadolinium-ethoxybenzyl-diethylenetriamine penta-acetic acid- (Gd-EOB-DTPA-) enhanced MRI. We developed a DCNN and deep fusion network (DFN) trained on full-size images, namely, DCNN-F and DFN-F. Furthermore, DFN was applied to the image blocks containing tumor lesions that were roughly contoured by a radiologist with 10 years of experience in abdominal MRI, and this method was named DFN-R. Another radiologist with five years of experience (moderate experience) performed tumor lesion contouring for comparison with our proposed methods. The ground truth image was contoured by an experienced radiologist and reviewed by an independent experienced radiologist. Results: The mean DSC of DCNN-F, DFN-F, and DFN-R was 0.69 ± 0.20 (median, 0.72), 0.74 ± 0.21 (median, 0.77), and 0.83 ± 0.13 (median, 0.88), respectively. The mean DSC of the segmentation by the radiologist with moderate experience was 0.79 ± 0.11 (median, 0.83), which was lower than the performance of DFN-R. Conclusions: Deep learning using dual-phase MRI shows great potential for HCC lesion segmentation. The radiologist-aided semiautomated method (DFN-R) achieved improved performance compared to manual contouring by the radiologist with moderate experience, although the difference was not statistically significant.

Hybrid split-arc partial-field volumetric modulated arc therapy: an improved beam arrangement for linear accelerator-based hippocampal-avoidance whole brain radiation therapy.

Background: This technical note aims to verify the hippocampus and adjacent organs at risk (OARs) sparing ability of an improved beam arrangement, namely hybrid split-arc partial-field volumetric modulated arc therapy (VMAT) (Hsapf-VMAT) during whole brain radiation therapy (WBRT). Materials and methods: Computed tomography simulation images of 22 patients with brain metastases were retrieved in this retrospective planning study. The hippocampus was manually delineated according to the criterion of RTOG 0933. Plans delivering 30 Gy in 10 fractions were generated for each patient using split-arc partial-field VMAT (sapf-VMAT) and Hsapf-VMAT. The sapf-VMAT plans consisted of 4 arc fields of 179.9° each with reduced field size. The Hsapf-VMAT consisted of 4 arc fields similar to sapf-VMAT in addition to 2 lateral opposing static fields. Statistical comparisons between treatment plans of both techniques were performed using the paired t-test at 5% level significance. Results: The results demonstrated that Hsapf-VMAT can achieve superior dose sparing in hippocampus which is comparable to sapf-VMAT (p > 0.05). In both eyes, Hsapf-VMAT had significantly lower Dmean and Dmax compared to sapf-VMAT (p < 0.005). Decrease in Dmax of both lenses using Hsapf-VMAT (p < 0.005) were statistically significant when compared to sapf-VMAT. Hsapf-VMAT demonstrated significant reduction of Dmean and Dmedian to the optic nerves (p < 0.05). Whole brain planning target volume (PTV) coverage was not compromised in both techniques. Conclusion: The present study adopts a hybrid technique, namely Hsapf-VMAT, for hippocampal sparing WBRT. Hsapf-VMAT can achieve promising dose reduction to the hippocampus, both eyes and lenses. Therefore, Hsapf-VMAT can be considered an improved version of sapf-VMAT.

Volumetric modulated arc therapy (VMAT) for hippocampal-avoidance whole brain radiation therapy: planning comparison with Dual-arc and Split-arc partial-field techniques.

BACKGROUND: Although whole brain radiation therapy (WBRT) provides palliation and prophylaxis, reduces local recurrence probability and improves overall survival, it is evident that WBRT is associated with neurocognitive deficits due to radiation induced damage of the hippocampus. Therefore, minimizing hippocampal dose to the least possible level is of high clinical relevance. In dual-arc conventional volumetric modulated arc therapy (dac-VMAT), the large irradiation field for whole brain planned target volume (PTV) requires a wide jaw opening in which substantial low dose volume to the hippocampus may be produced due to suboptimal multi-leaf collimator (MLC) movements. The present study investigates the potential of a radiation therapy technique with split-arc and reduced field size, namely split-arc partial-field volumetric modulated arc therapy (sapf-VMAT) to spare the hippocampus during WBRT. METHODS: Computed tomography and magnetic resonance images of 20 patients with brain metastases were retrieved in this retrospective planning study. The hippocampus was manually delineated by single radiation oncologist strictly following the RTOG 0933 atlas definition. Plans delivering 30 Gy in 10 fractions were generated for each patient using dac-VMAT and sapf-VMAT. Dosimetric parameters from both techniques were compared by paired t-test. RESULTS: The results demonstrated that radiation dose to the hippocampus was significantly reduced using sapf-VMAT relative to dac-VMAT plans. sapf-VMAT (7.86Gy, p = 0.001) had significantly lowered average D100% compared to dac-VMAT (9.23 Gy). Decrease in hippocampus Dmax using sapf-VMAT (13.23 Gy, p = 0.001) was statistically significant when compared to dac-VMAT (16.33 Gy). The resulting mean dose to the hippocampus was 9.16 Gy for the for sapf-VMAT. Mean dose of sapf-VMAT was significantly lower than dac-VMAT (10.85 Gy, p < 0.05). In both eyes, sapf-VMAT demonstrated significantly lower Dmean compared to dac-VMAT (p < 0.05). Whole brain PTV coverage was not compromised in both techniques. CONCLUSION: sapf-VMAT has demonstrated significant dose reduction to the hippocampus and both eyes compared to dac-VMAT.

Fully Automated Delineation of Gross Tumor Volume for Head and Neck Cancer on PET-CT Using Deep Learning: A Dual-Center Study.

Purpose: In this study, we proposed an automated deep learning (DL) method for head and neck cancer (HNC) gross tumor volume (GTV) contouring on positron emission tomography-computed tomography (PET-CT) images. Materials and Methods: PET-CT images were collected from 22 newly diagnosed HNC patients, of whom 17 (Database 1) and 5 (Database 2) were from two centers, respectively. An oncologist and a radiologist decided the gold standard of GTV manually by consensus. We developed a deep convolutional neural network (DCNN) and trained the network based on the two-dimensional PET-CT images and the gold standard of GTV in the training dataset. We did two experiments: Experiment 1, with Database 1 only, and Experiment 2, with both Databases 1 and 2. In both Experiment 1 and Experiment 2, we evaluated the proposed method using a leave-one-out cross-validation strategy. We compared the median results in Experiment 2 (GTVa) with the performance of other methods in the literature and with the gold standard (GTVm). Results: A tumor segmentation task for a patient on coregistered PET-CT images took less than one minute. The dice similarity coefficient (DSC) of the proposed method in Experiment 1 and Experiment 2 was 0.481∼0.872 and 0.482∼0.868, respectively. The DSC of GTVa was better than that in previous studies. A high correlation was found between GTVa and GTVm (R = 0.99, P < 0.001). The median volume difference (%) between GTVm and GTVa was 10.9%. The median values of DSC, sensitivity, and precision of GTVa were 0.785, 0.764, and 0.789, respectively. Conclusion: A fully automatic GTV contouring method for HNC based on DCNN and PET-CT from dual centers has been successfully proposed with high accuracy and efficiency. Our proposed method is of help to the clinicians in HNC management.

Microstructural brain abnormalities of children of idiopathic generalized epilepsy with generalized tonic-clonic seizure: a voxel-based diffusional kurtosis imaging study.

PURPOSE: To investigate the diffusion abnormalities in the brain of children with idiopathic generalized epilepsy (IGE) with generalized tonic-clonic seizure (GTCS) by using diffusion kurtosis imaging (DKI). MATERIALS AND METHODS: Twenty-one IGE children with GTCS and 16 controls were recruited. DKI was performed and maps of radial diffusivity (λ⊥ ), axial diffusivity (λ// ), mean diffusivity (MD), fractional anisotropy (FA), radial kurtosis (K⊥ ), axial kurtosis (K// ) and mean kurtosis (MK) were calculated. Voxel-based analyses were employed to compare diffusion metrics in epilepsy versus the controls. RESULTS: In the case group, MD was found significantly higher in the right temporal lobe, the right occipital lobe, hippocampus, and some subcortical regions, while FA increased in bilateral supplementary motor area and the left superior frontal lobe (false discovery rate corrected P < 0.05). Analysis of λ⊥ and λ// showed that the increased MD was mainly due to the elevated λ// . Significantly decreased MK was also detected in bilateral temporo-occipital regions, the right hippocampus, the left insula, the left post-central area, and some subcortical regions (false discovery rate corrected P < 0.05). In most regions the changed MK were due to the decreased K// . CONCLUSION: The kurtosis parameters (K⊥ , K// , and MK) reflect different microstructural information in the IGE children with GTCS, and this support the value of DKI in studying children GTCS.

Diffusion abnormalities in temporal lobes of children with temporal lobe epilepsy: a preliminary diffusional kurtosis imaging study and comparison with diffusion tensor imaging.

In this preliminary study, we aimed to investigate the abnormalities of water diffusion in children with temporal lobe epilepsy (TLE). Eight children with unilateral TLE (according to electroencephalography, EEG) and eight age- and sex-matched controls were recruited. Diffusion tensor imaging (DTI)/diffusional kurtosis imaging (DKI) acquisitions were performed. Radial diffusivity (λ(⊥)), axial diffusivity (λ(∥)), mean diffusivity (MD) and fractional anisotropy (FA) maps were calculated for both DTI and DKI, and radial kurtosis (K(⊥)), axial kurtosis (K(∥)) and mean kurtosis (MK) maps were calculated for DKI only. Mann-Whitney test showed that, for white matter in the temporal lobe, DKI-derived λ(∥) , MD and K(∥) were significantly different in bilateral temporal lobes and EEG-abnormal and EEG-normal sides of the temporal lobe between patients and controls, whereas DTI showed no abnormalities. For gray matter, DKI detected significantly higher MD and MK in the same three comparisons, whereas DTI detected abnormalities only in the comparison between bilateral temporal lobes and between EEG-normal sides in cases and left-right matched sides in controls. No significant difference was observed between EEG-abnormal and EEG-normal sides in cases. These preliminary results indicate that DKI is more sensitive than DTI for the detection of diffusion abnormalities in the temporal lobes of children with TLE, even when EEG signals are normal. These findings pave the way for the application of DKI for in-depth studies on TLE in children.

Factors affecting risk of symptomatic temporal lobe necrosis: significance of fractional dose and treatment time.

PURPOSE: To study the factors affecting the risk of symptomatic temporal lobe necrosis after different fractionation schedules. METHODS AND MATERIALS: One thousand thirty-two patients with T1-2 nasopharyngeal carcinoma treated with radical radiotherapy in Hong Kong during 1990-1995 were studied. They were treated at four different centers with similar techniques but different fractionation schedules: 984 patients were given 1 fraction daily throughout (q.d.), and 48 patients were irradiated twice daily (b.i.d.) for part of the course. The median total dose was 62.5 Gy (range 50.4-71.2), dose per fraction was 2.5 Gy (range 1.6-4.2), and overall treatment time (OTT) was 44 days (range 29-70). In addition, 500 patients received supplementary doses for parapharyngeal extension, 113 received booster doses by brachytherapy, and 114 received sequential chemotherapy using cisplatin-based regimes. RESULTS: Altogether, 24 patients developed symptomatic temporal lobe necrosis: 18 from the q.d. group and 6 from the b.i.d. group. The 5-year actuarial incidence ranged from 0% (after 66 Gy in 33 fractions within 44 days) to 14% (after 71.2 Gy in 40 fractions within 35 days). Multivariate analyses showed that the risk was significantly affected by the fractional effect of the product of total dose and dose per fraction (hazard ratio [HR] = 1.04, 95% confidence interval [CI] 1.02-1.05), OTT (HR 0.88, 95% CI 0.80-0.97), and b.i.d. scheduling (HR 13, 95% CI 3-54). Repeating the analyses for patients treated with the q.d. schedules confirmed the independent significance of OTT in addition to the product of total dose and dose per fraction. CONCLUSION: The tentative results suggest that in addition to fractional dose, the OTT also had significant impact on the risk of temporal lobe necrosis, and b.i.d. scheduling increased the hazard further.