Improved vascular depiction and image quality through deep learning reconstruction of CT hepatic arteriography during transcatheter arterial chemoembolization.

PURPOSE: To evaluate the effect of deep learning reconstruction (DLR) on vascular depiction, tumor enhancement, and image quality of computed tomography hepatic arteriography (CTHA) images acquired during transcatheter arterial chemoembolization (TACE). METHODS: Institutional review board approval was obtained. Twenty-seven patients (18 men and 9 women, mean age, 75.7 years) who underwent CTHA immediately before TACE were enrolled. All images were reconstructed using three reconstruction algorithms: hybrid-iterative reconstruction (hybrid-IR), DLR with mild strength (DLR-M), and DLR with strong strength (DLR-S). Vascular depiction, tumor enhancement, feeder visualization, and image quality of CTHA were quantitatively and qualitatively assessed by two radiologists and compared between the three reconstruction algorithms. RESULTS: The mean signal-to-noise ratios (SNR) of sub-segmental arteries and sub-sub-segmental arteries, and the contrast-to-noise ratio (CNR) of tumors, were significantly higher on DLR-S than on DLR-M and hybrid-IR (P < 0.001). The mean qualitative score for sharpness of sub-segmental and sub-sub-segmental arteries was significantly better on DLR-S than on DLR-M and hybrid-IR (P < 0.001). There was no significant difference in the feeder artery detection rate of automated feeder artery detection software among three reconstruction algorithms (P = 0.102). The contrast, continuity, and confidence level of feeder artery detection was significantly better on DLR-S than on DLR-M (P = 0.013, 0.005, and 0.001) and hybrid-IR (P < 0.001, P = 0.002, and P < 0.001). The weighted kappa values between two readers for qualitative scores of feeder artery visualization were 0.807-0.874. The mean qualitative scores for sharpness, granulation, and diagnostic acceptability of CTHA were better on DLR-S than on DLR-M and hybrid-IR (P < 0.001). CONCLUSIONS: DLR significantly improved the SNR of small hepatic arteries, the CNR of tumor, and feeder artery visualization on CTHA images. DLR-S seems to be better suited to routine CTHA in TACE than does hybrid-IR.

Evaluation of the performance of both machine learning models using PET and CT radiomics for predicting recurrence following lung stereotactic body radiation therapy: A single-institutional study.

PURPOSE: Predicting recurrence following stereotactic body radiotherapy (SBRT) for non-small cell lung cancer provides important information for the feasibility of the individualized radiotherapy and allows to select the appropriate treatment strategy based on the risk of recurrence. In this study, we evaluated the performance of both machine learning models using positron emission tomography (PET) and computed tomography (CT) radiomic features for predicting recurrence after SBRT. METHODS: Planning CT and PET images of 82 non-small cell lung cancer patients who performed SBRT at our hospital were used. First, tumors were delineated on each CT and PET of each patient, and 111 unique radiomic features were extracted, respectively. Next, the 10 features were selected using three different feature selection algorithms, respectively. Recurrence prediction models based on the selected features and four different machine learning algorithms were developed, respectively. Finally, we compared the predictive performance of each model for each recurrence pattern using the mean area under the curve (AUC) calculated following the 0.632+ bootstrap method. RESULTS: The highest performance for local recurrence, regional lymph node metastasis, and distant metastasis were observed in models using Support vector machine with PET features (mean AUC = 0.646), Naive Bayes with PET features (mean AUC = 0.611), and Support vector machine with CT features (mean AUC = 0.645), respectively. CONCLUSIONS: We comprehensively evaluated the performance of prediction model developed for recurrence following SBRT. The model in this study would provide information to predict the recurrence pattern and assist in making treatment strategies.

Six-week oral prednisolone therapy for immune-related pneumonitis: a single-arm phase II study.

BACKGROUND: There has been no prospective trial for treatment of immune-related pneumonitis (irP) occurred after immune checkpoint inhibitors (ICIs). METHODS: In this single-arm phase II study, patients with cancer with grade ≥2 irP received oral prednisolone (1 mg/kg/day), tapered over 6 weeks. The primary endpoint was a pneumonitis control rate at 6 weeks from the start of the study treatment, defined as complete disappearance or partial improvement of irP in high-resolution CT of the chest. RESULTS: Among 57 patients enrolled, 56 were included in the final analysis. The most frequent cause of irP was single ICI therapy (51.8%), followed by combination with chemotherapy plus ICI (39.3%). Thirty-five (62.5%) patients had grade 2 irP and 21 (37.5%) had grade ≥3. Fifty-one (91.1%) patients completed the study treatment while 5 discontinued the study treatment because of relapse of irP (n=1), death from cancer (n=1), occurrence of immune-related hepatitis (n=1), extension of the treatment duration more than 6 weeks (n=1), and attending physician's decision (n=1). Six weeks after the start of the study treatment, 16 (28.5%) patients demonstrated complete recovery from irP, 35 (62.5%) had a partial improvement in irP, 1 (1.8%) had a relapse of irP, and 4 (7.1%) were not evaluable. The pneumonitis control rate at 6 weeks was 91.1% (95% CI, 80.7% to 96.1%). Twelve weeks after the start of the study treatment, 5 (8.9%), 27 (48.2%), and 15 (26.8%) patients demonstrated complete recovery, partial improvement, and relapse, respectively, and 9 (16.1%) were not evaluable. The pneumonitis control rate at 12 weeks was 57.1% (95% CI, 44.1% to 69.2%). During the observation period, 18 (32.1%) patients experienced a relapse of irP, and of those, 17 received re-treatment with corticosteroids. Grade ≥3 adverse events occurred in 10 (17.9%) patients, in which hyperglycemia was most frequent (n=6). There was no treatment-related death. CONCLUSIONS: In this first prospective study for irP, prednisolone at 1 mg/kg/day, tapered over 6 weeks, demonstrated a promising clinical benefit and manageable toxicity, suggesting a potential treatment option for irP. TRIAL REGISTRATION NUMBER: jRCT: 1041190029.

Semi-automated histogram analysis of normal bone marrow using 18F-FDG PET/CT: correlation with clinical indicators.

BACKGROUND: 18F-fluorodeoxyglucose (FDG) positron emission tomography (PET) is increasingly applied to the diagnosis of bone marrow failure such as myeloproliferative neoplasm, aplastic anemia, and myelodysplastic syndrome, as well as malignant lymphoma and multiple myeloma. However, few studies have shown a normal FDG uptake pattern. This study aimed to establish a standard of bone marrow FDG uptake by a reproducible quantitative method with fewer steps using deep learning-based organ segmentation. METHODS: Bone marrow PET images were obtained using segmented whole-spine and pelvic bone marrow cavity CT as mask images using a commercially available imaging workstation that implemented an automatic organ segmentation algorithm based on deep learning. The correlation between clinical indicators and quantitative PET parameters, including histogram features, was evaluated. RESULTS: A total of 98 healthy adults were analyzed. The volume of bone marrow PET extracted in men was significantly higher than that in women (p < 0.0001). Univariate and multivariate regression analyses showed that mean of standardized uptake value corrected by lean body mass (SULmean) and entropy in both men and women were inversely correlated with age (all p < 0.0001), and SULmax in women were also inversely correlated with age (p = 0.011). CONCLUSION: A normal FDG uptake pattern was demonstrated by simplified FDG PET/CT bone marrow quantification.

Modified Glasgow Prognostic Score is predictive of prognosis for non-small cell lung cancer patients treated with stereotactic body radiation therapy: a retrospective study.

We aimed to assess the predictive value of the modified Glasgow prognostic score (mGPS) in patients with non-small cell lung cancer (NSCLC) who underwent stereotactic body radiation therapy (SBRT). We retrospectively reviewed the records of 207 patients, with a median age of 79 years. The pretreatment mGPS was calculated and categorized as high (mGPS = 1-2) or low (mGPS = 0). The median follow-up duration was 40.7 months. The five-year overall survival (OS), progression-free survival (PFS) and time to progression (TTP) rates were 44.3%, 36% and 54.4%, respectively. Multivariate analysis revealed that mGPS was independently predictive of OS (hazard ratio [HR] 1.67; 95% confidence interval 1.14-2.44: P = 0.009), PFS (HR 1.58; 1.10-2.28: P = 0.014) and TTP (HR 1.66; 1.03-2.68: P = 0.039). Patients who had high mGPS showed significantly worse OS (33.3 vs 64.5 months, P = 0.003) and worse PFS (23.8 vs 39 months, P = 0.008) than those who had low mGPS. The data showed a trend that patients with high mGPS suffered earlier progression compared to those with low mGPS (54.3 vs 88.1 months, P = 0.149). We confirmed that mGPS is independently predictive of prognosis in NSCLC patients treated with SBRT.

Tumor volume shrinkage during stereotactic body radiotherapy is related to better prognoses in patients with stage I non-small-cell lung cancer.

The purpose of the study was to investigate the association between tumor volume changes during stereotactic body radiation therapy (SBRT) and prognoses in stage I non-small-cell lung cancer (NSCLC). This retrospective review included stage I NSCLC patients in whom SBRT was performed at a total dose of 48.0-50.5 Gy in four or five fractions. The tumor volumes observed on computed tomography (CT) simulation and on the CT performed at the last treatment session using a CT-on-rails system were measured and compared. Then, the tumor volume changes during the SBRT period were measured and assessed for their association with prognoses (overall survival, local control, lymph node metastases and distant metastases). A total of 98 patients with a mean age of 78.6 years were enrolled in the study. The T-stage was T1a in 42%, T1b in 32% and T2a in 26% of the cases. The gross tumor volume (GTV) shrank and increased ≥10% in 23 (23.5%) and 36 (36.7%) of the cases, respectively. The 5-year local control and overall survival rates in the groups with a tumor shrinkage of ≥10% vs the group with a shrinkage of <10% were 94.7 vs 70.8% and 85.4 vs 47.6%, respectively; these differences were significant, with a P-value < 0.05. During a short SBRT period, the tumor shrank or enlarged in a small number of cases. A decrease of ≥10% in the GTV during SBRT was significantly related to better overall survival and local control.

Clinical Evaluation of Respiratory-triggered 3D MRCP with Navigator Echoes Compared to Breath-hold Acquisition Using Compressed Sensing and/or Parallel Imaging.

PURPOSE: To compare the image quality of three-dimensional magnetic resonance cholangiopancreatography (MRCP) acquired with respiratory triggering against breath-hold 3D MRCP with compressed sensing (CS) and parallel imaging (PI) in a clinical setting. METHODS: This study included 93 patients (45 men, mean age: 69.7 ± 9.3 years), in whom three types of 3D MRCP were performed: 3D breath-hold MRCP with CS and PI reconstruction (BH-CS-MRCP) and PI only reconstruction (BH-PI-MRCP) additionally to 3D respiratory triggered MRCP with navigator echoes (Nav-MRCP). Duct visualization and overall image quality were blindly evaluated on a four-point scale by two independent radiologists. Quantitative analysis was performed by calculating the relative duct-to-periductal contrast (RC) of three main biliary segments. Comparison between the methods was performed using paired t-test. RESULTS: Acquisition time was 23 s for both breath-hold MRCP protocols and 1 min 29 s for Nav-MRCP. Mean grading (Nav/CS/PI) for common bile duct (2.74/2.87/2.94), common hepatic duct (2.82/2.92/3.00), central right hepatic duct (2.75/2.85/2.98), central left hepatic duct (2.75/2.85/2.92) and cystic duct (2.22/2.34/2.42) was higher in BH-CS- and BH-PI-MRCP, whereas Nav-MRCP showed higher grading in the peripheral segments (peripheral right hepatic duct: 2.24/2.01/2.12; peripheral left hepatic duct: 2.23/2.02/2.13). Overall image quality of Nav-MRCP (2.91 ± 0.7) was not different from BH-PI-MRCP (2.92 ± 0.6) (P = 0.163), but higher than BH-CS-MRCP (2.80 ± 0.7) (P = 0.031). Quantitative analysis showed lower RC values for CS- and PI-MRCP than Nav-MRCP (P < 0.001). CONCLUSION: Breath-hold 3D MRCP were feasible using PI and CS. Visualization of the greater ductal system was even superior in breath-hold MRCP than in Nav-MRCP by considerably reducing acquisition time. Both breath-hold methods are suitable for revised MRI protocols notably in patients with irregular respiratory cycle.

Texture Analysis in the Diagnosis of Primary Breast Cancer: Comparison of High-Resolution Dedicated Breast Positron Emission Tomography (dbPET) and Whole-Body PET/CT.

Objective: This retrospective study aimed to compare the ability to classify tumor characteristics of breast cancer (BC) of positron emission tomography (PET)-derived texture features between dedicated breast PET (dbPET) and whole-body PET/computed tomography (CT). Methods: Forty-four BCs scanned by both high-resolution ring-shaped dbPET and whole-body PET/CT were analyzed. The primary BC was extracted with a standardized uptake value (SUV) threshold segmentation method. On both dbPET and PET/CT images, 38 texture features were computed; their ability to classify tumor characteristics such as tumor (T)-category, lymph node (N)-category, molecular subtype, and Ki67 levels was compared. The texture features were evaluated using univariate and multivariate analyses following principal component analysis (PCA). AUC values were used to evaluate the diagnostic power of the computed texture features to classify BC characteristics. Results: Some texture features of dbPET and PET/CT were different between Tis-1 and T2-4 and between Luminal A and other groups, respectively. No association with texture features was found in the N-category or Ki67 level. In contrast, receiver-operating characteristic analysis using texture features' principal components showed that the AUC for classification of any BC characteristics were equally good for both dbPET and whole-body PET/CT. Conclusions: PET-based texture analysis of dbPET and whole-body PET/CT may have equally good classification power for BC.

Renal Cancer is Not Radioresistant: Slowly but Continuing Shrinkage of the Tumor After Stereotactic Body Radiation Therapy.

PURPOSE: To evaluate the safety and efficacy of stereotactic body radiation therapy for primary lesion of renal cell carcinoma with long-term and regular follow-up of tumor size and renal function. METHODS: This prospective study included 13 patients treated with stereotactic body radiation therapy for primary lesion of stage I renal cell carcinoma between August 2007 and June 2016 in our institution. Diagnosis of renal cell carcinoma was made by 2 radiologists using computed tomography or magnetic resonance imaging. A dosage of 60 Gy in 10 fractions or 70 Gy in 10 fractions was prescribed. The higher dose was selected if dose constraints were satisfied. Tumor response on imaging examination, local progression-free rate, overall survival, and toxicity were assessed. RESULTS: The mean follow-up period was 48.3 months (range: 11-108 months). The tumors showed very slow but continuous response during long-term follow-up. Three cases (23.1%) showed transient progression during the short follow-up. The mean duration until the day on which partial response was confirmed among the partial or complete response cases was 22.6 months (95% confidence interval, 15.3-30.0 months). Local progression-free rate was 92.3% for 3 years and overall survival rate 91.7% for 2 years and 71.3% for 3 years. Twelve cases (92.3%) had impaired renal function at baseline. Renal function decreased slowly and mildly in most of the cases, but 2 cases of solitary kidney showed grade 4 or 5 renal dysfunction. CONCLUSION: All renal tumors decreased in size slowly but continuously for years after stereotactic body radiation therapy. Renal cancer can be treated radically with stereotactic body radiation therapy as a radiosensitive tumor, but careful attention should be given in cases with solitary kidney.

Magnetic resonance elastography is as accurate as liver biopsy for liver fibrosis staging.

BACKGROUND: Liver MR elastography (MRE) is available for the noninvasive assessment of liver fibrosis; however, no previous studies have compared the diagnostic ability of MRE with that of liver biopsy. PURPOSE: To compare the diagnostic accuracy of liver fibrosis staging between MRE-based methods and liver biopsy using the resected liver specimens as the reference standard. STUDY TYPE: A retrospective study at a single institution. POPULATION: In all, 200 patients who underwent preoperative MRE and subsequent surgical liver resection were included in this study. Data from 80 patients were used to estimate cutoff and distributions of liver stiffness values measured by MRE for each liver fibrosis stage (F0-F4, METAVIR system). In the remaining 120 patients, liver biopsy specimens were obtained from the resected liver tissues using a standard biopsy needle. FIELD STRENGTH/SEQUENCE: 2D liver MRE with gradient-echo based sequence on a 1.5 or 3T scanner was used. ASSESSMENT: Two radiologists independently measured the liver stiffness value on MRE and two types of MRE-based methods (threshold and Bayesian prediction method) were applied. Two pathologists evaluated all biopsy samples independently to stage liver fibrosis. Surgically resected whole tissue specimens were used as the reference standard. STATISTICAL TESTS: The accuracy for liver fibrosis staging was compared between liver biopsy and MRE-based methods with a modified McNemar's test. RESULTS: Accurate fibrosis staging was achieved in 53.3% (64/120) and 59.1% (71/120) of patients using MRE with threshold and Bayesian methods, respectively, and in 51.6% (62/120) with liver biopsy. Accuracies of MRE-based methods for diagnoses of ≥F2 (90-91% [108-9/120]), ≥F3 (79-81% [95-97/120]), and F4 (82-85% [98-102/120]) were statistically equivalent to those of liver biopsy (≥F2, 79% [95/120], P ≤ 0.01; ≥F3, 88% [105/120], P ≤ 0.006; and F4, 82% [99/120], P ≤ 0.017). DATA CONCLUSION: MRE can be an alternative to liver biopsy for fibrosis staging. LEVEL OF EVIDENCE: 3. Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;47:1268-1275.

Construction of mass spectra database and diagnosis algorithm for head and neck squamous cell carcinoma.

OBJECTIVES: Intraoperative identification of tumor margins is essential to achieving complete tumor resection. However, the process of intraoperative pathological diagnosis involves cumbersome procedures, such as preparation of cryosections and microscopic examination, thus requiring more than 30 min. Moreover, intraoperative diagnoses made by examining cryosections are occasionally inconsistent with postoperative diagnoses made by examining paraffin-embedded sections because the former are of poorer quality. We sought to establish a more rapid accurate method of intraoperative assessment. MATERIALS AND METHODS: A diagnostic algorithm of head and neck squamous cell carcinoma (HNSCC) using machine learning was constructed by mass spectra obtained from 15 non-cancerous and 19 HNSCC specimens by probe electrospray ionization mass spectrometry (PESI-MS). The clinical validity of this system was evaluated using intraoperative specimens of HNSCC and normal mucosa. RESULTS: A total of 114 and 141 mass spectra were acquired from non-cancerous and cancerous specimens, respectively, using both positive- and negative-ion modes of PESI-MS. These data were fed into partial least squares-logistic regression (PLS-LR) to discriminate tumor-specific spectral patterns. Leave-one-patient-out cross validation of this algorithm in positive- and negative-ion modes showed accuracies in HNSCC diagnosis of 90.48% and 95.35%, respectively. In intraoperative specimens of HNSCC, this algorithm precisely defined the borders of the cancerous regions; these corresponded with those determined by examining histologic sections. The procedure took approximately 5 min. CONCLUSION: This diagnostic system, based on machine learning, enables accurate discrimination of cancerous regions and has the potential to provide rapid intraoperative assessment of HNSCC margins.