The effect of hormone therapy on physiological uptake of the endometrium on [18F]F-FDG PET in postmenopausal women.

OBJECTIVE: The effects of hormonal therapy, estrogen-based hormone replacement therapy (HRT), and anti-tumor hormone therapy, such as tamoxifen, on the physiological uptake of the endometrium on 2-deoxy-2[18F]fluoro-D-glucose ([18F]F-FDG) positron emission tomography (PET) in postmenopausal women have not been determined. We explored the effect of hormone therapy, particularly HRT, on physiological uptake in the endometrium of postmenopausal women. MATERIALS AND METHODS: Postmenopausal women receiving hormone therapy who underwent cancer screening using PET/computed tomography (CT) between June 2016 and April 2023 were included in the hormone therapy group (n = 21). Postmenopausal women with no history of hormone therapy were included in the control group (n = 49). First, the physiological endometrial uptake at menopausal age and at least 1 year thereafter was compared quantitatively (SUVmax) and qualitatively (4-point scale) in the control group, to assess when the endometrium ceased to show significant physiological [18F]F-FDG uptake after menopause. Endometrial uptake was compared between the hormone therapy and control groups. The association between HRT duration (months) and endometrial uptake (SUVmax) was evaluated. Endometrial thickness, measured using transvaginal ultrasonography, was also compared between the two groups. RESULTS: Endometrial uptake was significantly reduced both qualitatively and quantitatively (P < 0.05) at least 1 year after menopause in control patients, by which time most women (89.8%) no longer had significant endometrial uptake. The hormone therapy group (n = 21) showed higher FDG uptake in the endometrium compared to the control group (median SUVmax: 2.3 vs 1.9, P = 0.0011), as well as a higher visual score (P < 0.0001). HRT duration did not correlate with endometrial uptake (P = 0.097). Endometrial thickness in the hormone therapy group was significantly thicker than in the control group (median: 3.9 mm vs 1.8 mm, P = 0.002). CONCLUSION: Hormone therapy may affect physiological uptake in the endometrium in postmenopausal women.

Performance of dedicated breast PET in breast cancer screening: comparison with digital mammography plus digital breast tomosynthesis and ultrasound.

OBJECTIVE: To compare the diagnostic performance of dedicated breast positron emission tomography (dbPET) in breast cancer screening with digital mammography plus digital breast tomosynthesis (DM-DBT) and breast ultrasound (US). METHODS: Women who participated in opportunistic whole-body PET/computed tomography cancer screening programs with breast examinations using dbPET, DM-DBT, and US between 2016-2020, whose results were determined pathologically or by follow-up for at least 1 year, were included. DbPET, DM-DBT, and US assessments were classified into four diagnostic categories: A (no abnormality), B (mild abnormality), C (need for follow-up), and D (recommend further examination). Category D was defined as screening positive. Each modality's recall rate, sensitivity, specificity, and positive predictive value (PPV) were calculated per examination to evaluate their diagnostic performance for breast cancer. RESULTS: Out of 2156 screenings, 18 breast cancer cases were diagnosed during the follow-up period (10 invasive cancers and eight ductal carcinomas in situ [DCIS]). The recall rates for dbPET, DM-DBT, and US were 17.8%, 19.2%, and 9.4%, respectively. The recall rate of dbPET was highest in the first year and subsequently decreased to 11.4%. dbPET, DM-DBT, and US had sensitivities of 72.2%, 88.9%, and 83.3%; specificities of 82.6%, 81.4%, and 91.2%; and PPVs of 3.4%, 3.9%, and 7.4%, respectively. The sensitivities of dbPET, DM-DBT, and US for invasive cancers were 90%, 100%, and 90%, respectively. There were no significant differences between the modalities. One case of dbPET-false-negative invasive cancer was identified in retrospect. DbPET had 50% sensitivity for DCIS, while that of both DM-DBT and US was 75%. Furthermore, the specificity of dbPET in the first year was the lowest among all periods, and modalities increased over the years to 88.7%. The specificity of dbPET was significantly higher than that of DM-DBT (p < 0.01) in the last 3 years. CONCLUSIONS: DbPET had a compatible sensitivity to DM-DBT and breast US for invasive breast cancer. The specificity of dbPET was improved and became higher than that of DM-DBT. DbPET may be a feasible screening modality.

Deep learning-based algorithm improved radiologists' performance in bone metastases detection on CT.

OBJECTIVES: To develop and evaluate a deep learning-based algorithm (DLA) for automatic detection of bone metastases on CT. METHODS: This retrospective study included CT scans acquired at a single institution between 2009 and 2019. Positive scans with bone metastases and negative scans without bone metastasis were collected to train the DLA. Another 50 positive and 50 negative scans were collected separately from the training dataset and were divided into validation and test datasets at a 2:3 ratio. The clinical efficacy of the DLA was evaluated in an observer study with board-certified radiologists. Jackknife alternative free-response receiver operating characteristic analysis was used to evaluate observer performance. RESULTS: A total of 269 positive scans including 1375 bone metastases and 463 negative scans were collected for the training dataset. The number of lesions identified in the validation and test datasets was 49 and 75, respectively. The DLA achieved a sensitivity of 89.8% (44 of 49) with 0.775 false positives per case for the validation dataset and 82.7% (62 of 75) with 0.617 false positives per case for the test dataset. With the DLA, the overall performance of nine radiologists with reference to the weighted alternative free-response receiver operating characteristic figure of merit improved from 0.746 to 0.899 (p < .001). Furthermore, the mean interpretation time per case decreased from 168 to 85 s (p = .004). CONCLUSION: With the aid of the algorithm, the overall performance of radiologists in bone metastases detection improved, and the interpretation time decreased at the same time. KEY POINTS: • A deep learning-based algorithm for automatic detection of bone metastases on CT was developed. • In the observer study, overall performance of radiologists in bone metastases detection improved significantly with the aid of the algorithm. • Radiologists' interpretation time decreased at the same time.

Radiomic machine learning for pretreatment assessment of prognostic risk factors for endometrial cancer and its effects on radiologists' decisions of deep myometrial invasion.

PURPOSE: To evaluate radiomic machine learning (ML) classifiers based on multiparametric magnetic resonance images (MRI) in pretreatment assessment of endometrial cancer (EC) risk factors and to examine effects on radiologists' interpretation of deep myometrial invasion (dMI). METHODS: This retrospective study examined 200 consecutive patients with EC during January 2004 -March 2017, divided randomly to Discovery (n = 150) and Test (n = 50) datasets. Radiomic features of tumors were extracted from T2-weighted images, apparent diffusion coefficient map, and contrast enhanced T1-weighed images. Using the Discovery dataset, feature selection and hyperparameter tuning for XGBoost were performed. Ten classifiers were built to predict dMI, histological grade, lymphovascular invasion (LVI), and pelvic/paraaortic lymph node metastasis (PLNM/PALNM), respectively. Using the Test dataset, the diagnostic performances of ten classifiers were assessed by the area under the receiver operator characteristic curve (AUC). Next, four radiologists assessed dMI independently using MRI with a Likert scale before and after referring to inference of the ML classifier for the Test dataset. Then, AUCs obtained before and after reference were compared. RESULTS: In the Test dataset, mean AUC of ML classifiers for dMI, histological grade, LVI, PLNM, and PALNM were 0.83, 0.77, 0.81, 0.72, and 0.82. AUCs of all radiologists for dMI (0.83-0.88) were better than or equal to mean AUC of the ML classifier, which showed no statistically significant difference before and after the reference. CONCLUSION: Radiomic classifiers showed promise for pretreatment assessment of EC risk factors. Radiologists' inferences outperformed the ML classifier for dMI and showed no improvement by review.

Deep Learning Algorithm for Fully Automated Detection of Small (≤4 cm) Renal Cell Carcinoma in Contrast-Enhanced Computed Tomography Using a Multicenter Database.

OBJECTIVES: Renal cell carcinoma (RCC) is often found incidentally in asymptomatic individuals undergoing abdominal computed tomography (CT) examinations. The purpose of our study is to develop a deep learning-based algorithm for fully automated detection of small (≤4 cm) RCCs in contrast-enhanced CT images using a multicenter database and to evaluate its performance. MATERIALS AND METHODS: For the algorithmic detection of RCC, we retrospectively selected contrast-enhanced CT images of patients with histologically confirmed single RCC with a tumor diameter of 4 cm or less between January 2005 and May 2020 from 7 centers in the Japan Medical Image Database. A total of 453 patients from 6 centers were selected as dataset A, and 132 patients from 1 center were selected as dataset B. Dataset A was used for training and internal validation. Dataset B was used only for external validation. Nephrogenic phase images of multiphase CT or single-phase postcontrast CT images were used. Our algorithm consisted of 2-step segmentation models, kidney segmentation and tumor segmentation. For internal validation with dataset A, 10-fold cross-validation was applied. For external validation, the models trained with dataset A were tested on dataset B. The detection performance of the models was evaluated using accuracy, sensitivity, specificity, and the area under the curve (AUC). RESULTS: The mean ± SD diameters of RCCs in dataset A and dataset B were 2.67 ± 0.77 cm and 2.64 ± 0.78 cm, respectively. Our algorithm yielded an accuracy, sensitivity, and specificity of 88.3%, 84.3%, and 92.3%, respectively, with dataset A and 87.5%, 84.8%, and 90.2%, respectively, with dataset B. The AUC of the algorithm with dataset A and dataset B was 0.930 and 0.933, respectively. CONCLUSIONS: The proposed deep learning-based algorithm achieved high accuracy, sensitivity, specificity, and AUC for the detection of small RCCs with both internal and external validations, suggesting that this algorithm could contribute to the early detection of small RCCs.

Temporal subtraction CT with nonrigid image registration improves detection of bone metastases by radiologists: results of a large-scale observer study.

To determine whether temporal subtraction (TS) CT obtained with non-rigid image registration improves detection of various bone metastases during serial clinical follow-up examinations by numerous radiologists. Six board-certified radiologists retrospectively scrutinized CT images for patients with history of malignancy sequentially. These radiologists selected 50 positive and 50 negative subjects with and without bone metastases, respectively. Furthermore, for each subject, they selected a pair of previous and current CT images satisfying predefined criteria by consensus. Previous images were non-rigidly transformed to match current images and subtracted from current images to automatically generate TS images. Subsequently, 18 radiologists independently interpreted the 100 CT image pairs to identify bone metastases, both without and with TS images, with each interpretation separated from the other by an interval of at least 30 days. Jackknife free-response receiver operating characteristics (JAFROC) analysis was conducted to assess observer performance. Compared with interpretation without TS images, interpretation with TS images was associated with a significantly higher mean figure of merit (0.710 vs. 0.658; JAFROC analysis, P = 0.0027). Mean sensitivity at lesion-based was significantly higher for interpretation with TS compared with that without TS (46.1% vs. 33.9%; P = 0.003). Mean false positive count per subject was also significantly higher for interpretation with TS than for that without TS (0.28 vs. 0.15; P < 0.001). At the subject-based, mean sensitivity was significantly higher for interpretation with TS images than that without TS images (73.2% vs. 65.4%; P = 0.003). There was no significant difference in mean specificity (0.93 vs. 0.95; P = 0.083). TS significantly improved overall performance in the detection of various bone metastases.

A Proposed Dedicated Breast PET Lexicon: Standardization of Description and Reporting of Radiotracer Uptake in the Breast.

Dedicated breast positron emission tomography (dbPET) is a new diagnostic imaging modality recently used in clinical practice for the detection of breast cancer and the assessment of tumor biology. dbPET has higher spatial resolution than that of conventional whole body PET systems, allowing recognition of detailed morphological attributes of radiotracer accumulation within the breast. 18F-fluorodeoxyglucose (18F-FDG) accumulation in the breast may be due to benign or malignant entities, and recent studies suggest that morphology characterization of 18F-FDG uptake could aid in estimating the probability of malignancy. However, across the world, there are many descriptors of breast 18F-FDG uptake, limiting comparisons between studies. In this article, we propose a lexicon for breast radiotracer uptake to standardize description and reporting of image findings on dbPET, consisting of terms for image quality, radiotracer fibroglandular uptake, breast lesion uptake.

Automatic segmentation of uterine endometrial cancer on multi-sequence MRI using a convolutional neural network.

Endometrial cancer (EC) is the most common gynecological tumor in developed countries, and preoperative risk stratification is essential for personalized medicine. There have been several radiomics studies for noninvasive risk stratification of EC using MRI. Although tumor segmentation is usually necessary for these studies, manual segmentation is not only labor-intensive but may also be subjective. Therefore, our study aimed to perform the automatic segmentation of EC on MRI with a convolutional neural network. The effect of the input image sequence and batch size on the segmentation performance was also investigated. Of 200 patients with EC, 180 patients were used for training the modified U-net model; 20 patients for testing the segmentation performance and the robustness of automatically extracted radiomics features. Using multi-sequence images and larger batch size was effective for improving segmentation accuracy. The mean Dice similarity coefficient, sensitivity, and positive predictive value of our model for the test set were 0.806, 0.816, and 0.834, respectively. The robustness of automatically extracted first-order and shape-based features was high (median ICC = 0.86 and 0.96, respectively). Other high-order features presented moderate-high robustness (median ICC = 0.57-0.93). Our model could automatically segment EC on MRI and extract radiomics features with high reliability.

Visceral fat obesity is the key risk factor for the development of reflux erosive esophagitis in 40-69-years subjects.

BACKGROUND: Visceral fat obesity can be defined quantitatively by abdominal computed tomography, however, the usefulness of measuring visceral fat area to assess the etiology of gastrointestinal reflux disease has not been fully elucidated. METHODS: A total of 433 healthy subjects aged 40-69 years (234 men, 199 women) were included in the study. The relationship between obesity-related factors (total fat area, visceral fat area, subcutaneous fat area, waist circumference, and body mass index) and the incidence of reflux erosive esophagitis was investigated. Lifestyle factors and stomach conditions relevant to the onset of erosive esophagitis were also analyzed. RESULTS: The prevalence of reflux erosive esophagitis was 27.2% (118/433; 106 men, 12 women). Visceral fat area was higher in subjects with erosive esophagitis than in those without (116.6 cm2 vs. 64.9 cm2, respectively). The incidence of erosive esophagitis was higher in subjects with visceral fat obesity (visceral fat area ≥ 100 cm2) than in those without (61.2% vs. 12.8%, respectively). Visceral fat obesity had the highest odds ratio (OR) among obesity-related factors. Multivariate analysis showed that visceral fat area was associated with the incidence of erosive esophagitis (OR = 2.18), indicating that it is an independent risk factor for erosive esophagitis. In addition, daily alcohol intake (OR = 1.54), gastric atrophy open type (OR = 0.29), and never-smoking history (OR = 0.49) were also independently associated with the development of erosive esophagitis. CONCLUSIONS: Visceral fat obesity is the key risk factor for the development of reflux erosive esophagitis in subjects aged 40-69 years.

Adaptive Voxel Matching for Temporal CT Subtraction.

Temporal subtraction (TS) technique calculates a subtraction image between a pair of registered images acquired from the same patient at different times. Previous studies have shown that TS is effective for visualizing pathological changes over time; therefore, TS should be a useful tool for radiologists. However, artifacts caused by partial volume effects degrade the quality of thick-slice subtraction images, even with accurate image registration. Here, we propose a subtraction method for reducing artifacts in thick-slice images and discuss its implementation in high-speed processing. The proposed method is based on voxel matching, which reduces artifacts by considering gaps in discretized positions of two images in subtraction calculations. There are two different features between the proposed method and conventional voxel matching: (1) the size of a searching region to reduce artifacts is determined based on discretized position gaps between images and (2) the searching region is set on both images for symmetrical subtraction. The proposed method is implemented by adopting an accelerated subtraction calculation method that exploit the nature of liner interpolation for calculating the signal value at a point among discretized positions. We quantitatively evaluated the proposed method using synthetic data and qualitatively using clinical data interpreted by radiologists. The evaluation showed that the proposed method was superior to conventional methods. Moreover, the processing speed using the proposed method was almost unchanged from that of the conventional methods. The results indicate that the proposed method can improve the quality of subtraction images acquired from thick-slice images.

Bone segmentation on whole-body CT using convolutional neural network with novel data augmentation techniques.

BACKGROUND: The purpose of this study was to develop and evaluate an algorithm for bone segmentation on whole-body CT using a convolutional neural network (CNN). METHODS: Bone segmentation was performed using a network based on U-Net architecture. To evaluate its performance and robustness, we prepared three different datasets: (1) an in-house dataset comprising 16,218 slices of CT images from 32 scans in 16 patients; (2) a secondary dataset comprising 12,529 slices of CT images from 20 scans in 20 patients, which were collected from The Cancer Imaging Archive; and (3) a publicly available labelled dataset comprising 270 slices of CT images from 27 scans in 20 patients. To improve the network's performance and robustness, we evaluated the efficacy of three types of data augmentation technique: conventional method, mixup, and random image cropping and patching (RICAP). RESULTS: The network trained on the in-house dataset achieved a mean Dice coefficient of 0.983 ± 0.005 on cross validation with the in-house dataset, and 0.943 ± 0.007 with the secondary dataset. The network trained on the public dataset achieved a mean Dice coefficient of 0.947 ± 0.013 on 10 randomly generated 15-3-9 splits of the public dataset. These results outperform those reported previously. Regarding augmentation technique, the conventional method, RICAP, and a combination of these were effective. CONCLUSIONS: The CNN-based model achieved accurate bone segmentation on whole-body CT, with generalizability to various scan conditions. Data augmentation techniques enabled construction of an accurate and robust model even with a small dataset.

Improving breast mass classification by shared data with domain transformation using a generative adversarial network.

Training of a convolutional neural network (CNN) generally requires a large dataset. However, it is not easy to collect a large medical image dataset. The purpose of this study is to investigate the utility of synthetic images in training CNNs and to demonstrate the applicability of unrelated images by domain transformation. Mammograms showing 202 benign and 212 malignant masses were used for evaluation. To create synthetic data, a cycle generative adversarial network was trained with 599 lung nodules in computed tomography (CT) and 1430 breast masses on digitized mammograms (DDSM). A CNN was trained for classification between benign and malignant masses. The classification performance was compared between the networks trained with the original data, augmented data, synthetic data, DDSM images, and natural images (ImageNet dataset). The results were evaluated in terms of the classification accuracy and the area under the receiver operating characteristic curves (AUC). The classification accuracy improved from 65.7% to 67.1% with data augmentation. The use of an ImageNet pretrained model was useful (79.2%). Performance was slightly improved when synthetic images or the DDSM images only were used for pretraining (67.6 and 72.5%, respectively). When the ImageNet pretrained model was trained with the synthetic images, the classification performance slightly improved (81.4%), although the difference in AUCs was not statistically significant. The use of the synthetic images had an effect similar to the DDSM images. The results of the proposed study indicated that the synthetic data generated from unrelated lesions by domain transformation could be used to increase the training samples.

Automatic segmentation of the uterus on MRI using a convolutional neural network.

BACKGROUND: This study was performed to evaluate the clinical feasibility of a U-net for fully automatic uterine segmentation on MRI by using images of major uterine disorders. METHODS: This study included 122 female patients (14 with uterine endometrial cancer, 15 with uterine cervical cancer, and 55 with uterine leiomyoma). U-net architecture optimized for our research was used for automatic segmentation. Three-fold cross-validation was performed for validation. The results of manual segmentation of the uterus by a radiologist on T2-weighted sagittal images were used as the gold standard. Dice similarity coefficient (DSC) and mean absolute distance (MAD) were used for quantitative evaluation of the automatic segmentation. Visual evaluation using a 4-point scale was performed by two radiologists. DSC, MAD, and the score of the visual evaluation were compared between uteruses with and without uterine disorders. RESULTS: The mean DSC of our model for all patients was 0.82. The mean DSCs for patients with and without uterine disorders were 0.84 and 0.78, respectively (p = 0.19). The mean MADs for patients with and without uterine disorders were 18.5 and 21.4 [pixels], respectively (p = 0.39). The scores of the visual evaluation were not significantly different between uteruses with and without uterine disorders. CONCLUSIONS: Fully automatic uterine segmentation with our modified U-net was clinically feasible. The performance of the segmentation of our model was not influenced by the presence of uterine disorders.

Temporal subtraction of computed tomography images improves detectability of bone metastases by radiology residents.

OBJECTIVE: Temporal subtraction of CT (TS) images improves detection of newly developed bone metastases (BM). We sought to determine whether TS improves detection of BM by radiology residents as well. METHODS: We performed an observer study using a previously reported dataset, consisting of 60 oncology patients, each with previous and current CT images. TS images were calculated using in-house software. Four residents independently interpreted twice the 60 sets of CT images, without and with TS. They identified BM by marking suspicious lesions likely to be BM. Lesion-based sensitivity and number of false positives per patient were calculated. Figure-of-merit (FOM) was calculated. Detectability of BM, with and without TS, was compared between radiology residents and board-certified radiologists, as published previously. RESULTS: FOM of residents significantly improved by implementing TS (p value < 0.0001). Lesion-based sensitivity, false positives per patients, and FOM were 40.8%, 0.121, and 0.657, respectively, without TS, and 58.1%, 0.0958, and 0.796, respectively, with TS. These findings were comparable with the previously published values for board-certified radiologists without TS (58.0%, 0.19, and 0.758, respectively). CONCLUSION: The detectability of BM by residents improved markedly by implementing TS and reached that of board-certified radiologists without TS. KEY POINTS: • Detectability of bone metastases on CT by residents improved significantly when using temporal subtraction of CT (TS). • Detections by residents with TS and board-certified radiologists without TS were comparable. • TS is useful for residents as it is for board-certified radiologists.

CT temporal subtraction improves early detection of bone metastases compared to SPECT.

OBJECTIVES: To compare observer performance of detecting bone metastases between bone scintigraphy, including planar scan and single-photon emission computed tomography, and computed tomography (CT) temporal subtraction (TS). METHODS: Data on 60 patients with cancer who had undergone CT (previous and current) and bone scintigraphy were collected. Previous CT images were registered to the current ones by large deformation diffeomorphic metric mapping; the registered previous images were subtracted from the current ones to produce TS. Definitive diagnosis of bone metastases was determined by consensus between two radiologists. Twelve readers independently interpreted the following pairs of examinations: NM-pair, previous and current CTs and bone scintigraphy, and TS-pair, previous and current CTs and TS. The readers assigned likelihood levels to suspected bone metastases for diagnosis. Sensitivity, number of false positives per patient (FPP), and reading time for each pair of examinations were analysed for evaluating observer performance by performing the Wilcoxon signed-rank test. Figure-of-merit (FOM) was calculated using jackknife alternative free-response receiver operating characteristic analysis. RESULTS: The sensitivity of TS was significantly higher than that of bone scintigraphy (54.3% vs. 41.3%, p = 0.006). FPP with TS was significantly higher than that with bone scintigraphy (0.189 vs. 0.0722, p = 0.003). FOM of TS tended to be better than that of bone scintigraphy (0.742 vs. 0.691, p = 0.070). CONCLUSION: Sensitivity of TS in detecting bone metastasis was significantly higher than that of bone scintigraphy, but still limited to 54%. TS might be superior to bone scintigraphy for early detection of bone metastasis. KEY POINTS: • Computed tomography temporal subtraction was helpful in early detection of bone metastases. • Sensitivity for bone metastasis was higher for computed tomography temporal subtraction than for bone scintigraphy. • Figure-of-merit of computed tomography temporal subtraction was better than that of bone scintigraphy.

Detection of suspected brain infarctions on CT can be significantly improved with temporal subtraction images.

OBJECTIVE: To assess whether temporal subtraction (TS) images of brain CT improve the detection of suspected brain infarctions. METHODS: Study protocols were approved by our institutional review board, and informed consent was waived because of the retrospective nature of this study. Forty-two sets of brain CT images of 41 patients, each consisting of a pair of brain CT images scanned at two time points (previous and current) between January 2011 and November 2016, were collected for an observer performance study. The 42 sets consisted of 23 cases with a total of 77 newly developed brain infarcts or hyperdense artery signs confirmed by two radiologists who referred to additional clinical information and 19 negative control cases. To create TS images, the previous images were registered to the current images by partly using a non-rigid registration algorithm and then subtracted. Fourteen radiologists independently interpreted the images to identify the lesions with and without TS images with an interval of over 4 weeks. A figure of merit (FOM) was calculated along with the jackknife alternative free-response receiver-operating characteristic analysis. Sensitivity, number of false positives per case (FPC) and reading time were analyzed by the Wilcoxon signed-rank test. RESULTS: The mean FOM increased from 0.528 to 0.737 with TS images (p < 0.0001). The mean sensitivity and FPC improved from 26.5% and 0.243 to 56.0% and 0.153 (p < 0.0001 and p = 0.239), respectively. The mean reading time was 173 s without TS and 170 s with TS (p = 0.925). CONCLUSION: The detectability of suspected brain infarctions was significantly improved with TS CT images. KEY POINTS: • Although it is established that MRI is superior to CT in the detection of strokes, the first choice of modality for suspected stroke patients is often CT. • An observer performance study with 14 radiologists was performed to evaluate whether temporal subtraction images derived from a non-rigid transformation algorithm can significantly improve the detectability of newly developed brain infarcts on CT. • Temporal subtraction images were shown to significantly improve the detectability of newly developed brain infarcts on CT.

Automatic inference model construction for computer-aided diagnosis of lung nodule: Explanation adequacy, inference accuracy, and experts' knowledge.

We aimed to describe the development of an inference model for computer-aided diagnosis of lung nodules that could provide valid reasoning for any inferences, thereby improving the interpretability and performance of the system. An automatic construction method was used that considered explanation adequacy and inference accuracy. In addition, we evaluated the usefulness of prior experts' (radiologists') knowledge while constructing the models. In total, 179 patients with lung nodules were included and divided into 79 and 100 cases for training and test data, respectively. F-measure and accuracy were used to assess explanation adequacy and inference accuracy, respectively. For F-measure, reasons were defined as proper subsets of Evidence that had a strong influence on the inference result. The inference models were automatically constructed using the Bayesian network and Markov chain Monte Carlo methods, selecting only those models that met the predefined criteria. During model constructions, we examined the effect of including radiologist's knowledge in the initial Bayesian network models. Performance of the best models in terms of F-measure, accuracy, and evaluation metric were as follows: 0.411, 72.0%, and 0.566, respectively, with prior knowledge, and 0.274, 65.0%, and 0.462, respectively, without prior knowledge. The best models with prior knowledge were then subjectively and independently evaluated by two radiologists using a 5-point scale, with 5, 3, and 1 representing beneficial, appropriate, and detrimental, respectively. The average scores by the two radiologists were 3.97 and 3.76 for the test data, indicating that the proposed computer-aided diagnosis system was acceptable to them. In conclusion, the proposed method incorporating radiologists' knowledge could help in eliminating radiologists' distrust of computer-aided diagnosis and improving its performance.

Computer-aided diagnosis of lung nodule classification between benign nodule, primary lung cancer, and metastatic lung cancer at different image size using deep convolutional neural network with transfer learning.

We developed a computer-aided diagnosis (CADx) method for classification between benign nodule, primary lung cancer, and metastatic lung cancer and evaluated the following: (i) the usefulness of the deep convolutional neural network (DCNN) for CADx of the ternary classification, compared with a conventional method (hand-crafted imaging feature plus machine learning), (ii) the effectiveness of transfer learning, and (iii) the effect of image size as the DCNN input. Among 1240 patients of previously-built database, computed tomography images and clinical information of 1236 patients were included. For the conventional method, CADx was performed by using rotation-invariant uniform-pattern local binary pattern on three orthogonal planes with a support vector machine. For the DCNN method, CADx was evaluated using the VGG-16 convolutional neural network with and without transfer learning, and hyperparameter optimization of the DCNN method was performed by random search. The best averaged validation accuracies of CADx were 55.9%, 68.0%, and 62.4% for the conventional method, the DCNN method with transfer learning, and the DCNN method without transfer learning, respectively. For image size of 56, 112, and 224, the best averaged validation accuracy for the DCNN with transfer learning were 60.7%, 64.7%, and 68.0%, respectively. DCNN was better than the conventional method for CADx, and the accuracy of DCNN improved when using transfer learning. Also, we found that larger image sizes as inputs to DCNN improved the accuracy of lung nodule classification.

Computer-aided diagnosis of lung nodule using gradient tree boosting and Bayesian optimization.

We aimed to evaluate a computer-aided diagnosis (CADx) system for lung nodule classification focussing on (i) usefulness of the conventional CADx system (hand-crafted imaging feature + machine learning algorithm), (ii) comparison between support vector machine (SVM) and gradient tree boosting (XGBoost) as machine learning algorithms, and (iii) effectiveness of parameter optimization using Bayesian optimization and random search. Data on 99 lung nodules (62 lung cancers and 37 benign lung nodules) were included from public databases of CT images. A variant of the local binary pattern was used for calculating a feature vector. SVM or XGBoost was trained using the feature vector and its corresponding label. Tree Parzen Estimator (TPE) was used as Bayesian optimization for parameters of SVM and XGBoost. Random search was done for comparison with TPE. Leave-one-out cross-validation was used for optimizing and evaluating the performance of our CADx system. Performance was evaluated using area under the curve (AUC) of receiver operating characteristic analysis. AUC was calculated 10 times, and its average was obtained. The best averaged AUC of SVM and XGBoost was 0.850 and 0.896, respectively; both were obtained using TPE. XGBoost was generally superior to SVM. Optimal parameters for achieving high AUC were obtained with fewer numbers of trials when using TPE, compared with random search. Bayesian optimization of SVM and XGBoost parameters was more efficient than random search. Based on observer study, AUC values of two board-certified radiologists were 0.898 and 0.822. The results show that diagnostic accuracy of our CADx system was comparable to that of radiologists with respect to classifying lung nodules.