Computer-assisted Detection Software Research Articles

Assessing Detection Accuracy of Computerized Sonographic Features and Computer-Assisted Reading Performance in Differentiating Thyroid Cancers.

For ultrasound imaging of thyroid nodules, medical guidelines are all based on findings of sonographic features to provide clinicians management recommendations. Due to the recent development of artificial intelligence and machine learning (AI/ML) technologies, there have been computer-assisted detection (CAD) software devices available for clinical use to detect and quantify the sonographic features of thyroid nodules. This study is to validate the accuracy of the computerized sonographic features (CSF) by a CAD software device, namely, AmCAD-UT, and then to assess how the reading performance of clinicians (readers) can be improved providing the computerized features. The feature detection accuracy is tested against the ground truth established by a panel of thyroid specialists and a multiple-reader multiple-case (MRMC) study is performed to assess the sequential reading performance with the assistance of the CSF. Five computerized features, including anechoic area, hyperechoic foci, hypoechoic pattern, heterogeneous texture, and indistinct margin, were tested, with AUCs ranging from 0.888~0.946, 0.825~0.913, 0.812~0.847, 0.627~0.77, and 0.676~0.766, respectively. With the five CSFs, the sequential reading performance of 18 clinicians is found significantly improved, with the AUC increasing from 0.720 without CSF to 0.776 with CSF. Our studies show that the computerized features are consistent with the clinicians’ findings and provide additional value in assisting sonographic diagnosis.

An Initiative to Reduce Unnecessary Gadolinium-Based Contrast in Multiple Sclerosis Patients

ObjectivePatients with multiple sclerosis (MS) routinely undergo serial contrast-enhanced MRIs. Given concerns regarding tissue deposition of gadolinium-based contrast agents (GBCAs) and evidence that enhancement of lesions is only seen in patients with new disease activity on noncontrast imaging, we set out to implement a prospective quality improvement project whereby intravenous contrast would be reserved only for patients with evidence of new disease activity on noncontrast images. MethodsTo prospectively implement such a protocol, we leveraged our in-house computer-assisted detection (CAD) software and 3-D laboratory radiology technologists to perform real-time preliminary assessments of the CAD-processed T2 fluid attenuated inversion recovery (FLAIR) noncontrast images as a basis for deciding whether to inject contrast. Before implementation, we held multidisciplinary meetings with neurology, neuroradiology, and MR technologists and distributed surveys to objectively assess opinions and obstacles to clinical implementation. We evaluated reduction in GBCA utilization and technologist performance relative to final neuroradiologist interpretations. ResultsDuring a 2-month trial period, 153 patients were imaged under the new protocol. Technologists using the CAD software were able to identify patients with new or enlarging lesions on FLAIR images with 95% accuracy and 97% negative predictive value relative to final neuroradiologist interpretations, which allowed us to avoid the use of contrast and additional imaging sequences in 87% of patients. DiscussionA multidisciplinary effort to implement a quality improvement project to limit contrast in MS patients receiving follow-up MRIs allowed for improved safety and cost by targeting patients that would benefit from the use of intravenous contrast in real-time.

Four-dimensional fully convolutional residual network-based liver segmentation in Gd-EOB-DTPA-enhanced MRI.

Gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) tends to show higher diagnostic accuracy than other modalities. There is a demand for computer-assisted detection (CAD) software for Gd-EOB-DTPA-enhanced MRI. Segmentation with high accuracy is important for CAD software. We propose a liver segmentation method for Gd-EOB-DTPA-enhanced MRI that is based on a four-dimensional (4D) fully convolutional residual network (FC-ResNet). The aims of this study are to determine the best combination of an input image and output image in our proposed method and to compare our proposed method with the previous rule-based segmentation method. We prepared a five-phase image set and a hepatobiliary phase image set as the input image sets to determine the best input image set. We also prepared a labeled liver image and labeled liver and labeled body trunk images as the output image sets to determine the best output image set. In addition, we optimized the hyperparameters of our proposed model. We used 30 cases to train our model, 10 cases to determine the hyperparameters of our model, and 20 cases to evaluate our model. Our network with the five-phase image set and the output image set of labeled liver and labeled body trunk images showed the highest accuracy. Our proposed method showed higher accuracy than the previous rule-based segmentation method. The Dice coefficient of the liver region was 0.944 ± 0.018. Our proposed 4D FC-ResNet showed satisfactory performance for liver segmentation as preprocessing in CAD software.

Can the spherical gold standards be used as an alternative to painted gold standards for the computerized detection of lesions using voxel-based classification?

For the development of computer-assisted detection (CAD) software using voxel-based classification, gold standards defined by pixel-by-pixel painting, called painted gold standards, are desirable. However, for radiologists who define gold standards, a simplified method of definition is desirable. One of the simplest methods of defining gold standards is a spherical region, called a spherical gold standard. In this study, we investigated whether spherical gold standards can be used as an alternative to painted gold standards for computerized detection using voxel-based classification. The spherical gold standards were determined by the center of gravity and the maximum diameter. We compared two types of gold standard, painted gold standards and spherical gold standards, by two types of CAD software using voxel-based classification. The time required to paint the area of one lesion was 4.7-6.5 times longer than the time required to define a spherical gold standard. For the same performance of the CAD software, the number of training cases required for the spherical gold standard was 1.6-7.6 times that for the painted gold standards. Spherical gold standards can be used as an alternative to painted gold standards for the computerized detection of lesions with simple shapes.

Effects of Iterative Reconstruction Algorithms on Computer-assisted Detection (CAD) Software for Lung Nodules in Ultra-low-dose CT for Lung Cancer Screening

This study aimed to evaluate the effects of iterative reconstruction (IR) algorithms on computer-assisted detection (CAD) software for lung nodules in ultra-low-dose computed tomography (ULD-CT) for lung cancer screening. We selected 85 subjects who underwent both a low-dose CT (LD-CT) scan and an additional ULD-CT scan in our lung cancer screening program for high-risk populations. The LD-CT scans were reconstructed with filtered back projection (FBP; LD-FBP). The ULD-CT scans were reconstructed with FBP (ULD-FBP), adaptive iterative dose reduction 3D (AIDR 3D; ULD-AIDR 3D), and forward projected model-based IR solution (FIRST; ULD-FIRST). CAD software for lung nodules was applied to each image dataset, and the performance of the CAD software was compared among the different IR algorithms. The mean volume CT dose indexes were 3.02 mGy (LD-CT) and 0.30 mGy (ULD-CT). For overall nodules, the sensitivities of CAD software at 3.0 false positives per case were 78.7% (LD-FBP), 9.3% (ULD-FBP), 69.4% (ULD-AIDR 3D), and 77.8% (ULD-FIRST). Statistical analysis showed that the sensitivities of ULD-AIDR 3D and ULD-FIRST were significantly higher than that of ULD-FBP (P < .001). The performance of CAD software in ULD-CT was improved by using IR algorithms. In particular, the performance of CAD in ULD-FIRST was almost equivalent to that in LD-FBP.

Performance of ultralow-dose CT with iterative reconstruction in lung cancer screening: limiting radiation exposure to the equivalent of conventional chest X-ray imaging.

To investigate the detection rate of pulmonary nodules in ultralow-dose CT acquisitions. In this lung phantom study, 232 nodules (115 solid, 117 ground-glass) of different sizes were randomly distributed in a lung phantom in 60 different arrangements. Every arrangement was acquired once with standard radiation dose (100 kVp, 100 references mAs) and once with ultralow radiation dose (80 kVp, 6 mAs). Iterative reconstruction was used with optimized kernels: I30 for ultralow-dose, I70 for standard dose and I50 for CAD. Six radiologists examined the axial 1-mm stack for solid and ground-glass nodules. During a second and third step, three radiologists used maximum intensity projection (MIPs), finally checking with computer-assisted detection (CAD), while the others first used CAD, finally checking with the MIPs. The detection rate was 95.5% with standard dose (DLP 126mGy*cm) and 93.3% with ultralow-dose (DLP: 9mGy*cm). The additional use of either MIP reconstructions or CAD software could compensate for this difference. A combination of both MIP reconstructions and CAD software resulted in a maximum detection rate of 97.5% with ultralow-dose. Lung cancer screening with ultralow-dose CT using the same radiation dose as a conventional chest X-ray is feasible. • 93.3% of all lung nodules were detected with ultralow-dose CT. • A sensitivity of 97.5 % is possible with additional image post-processing. • The radiation dose is comparable to a standard radiography in two planes. • Lung cancer screening with ultralow-dose CT is feasible.

Computed tomography with computer-assisted detection of pulmonary nodules in dogs and cats

The aim of this study was to assess the potential benefit of computer-assisted detection (CAD) of pulmonary nodules in veterinary medicine. Therefore, the CAD rate was compared to the detection rates of two individual examiners in terms of its sensitivity and false-positive findings. We included 51 dogs and 16 cats with pulmonary nodules previously diagnosed by computed tomography. First, the number of nodules ≥3mm was recorded for each patient by two independent examiners. Subsequently, each examiner used the CAD software for automated nodule detection. With the knowledge of the CAD results, a final consensus decision on the number of nodules was achieved. The software used was a commercially available CAD program. The sensitivity of examiner 1 was 89.2%, while that of examiner 2 reached 87.4%. CAD had a sensitivity of 69.4%. With CAD, the sensitivity of examiner 1 increased to 94.7% and that of examiner 2 to 90.8%. The CAD-system, which we used in our study, had a moderate sensitivity of 69.4%. Despite its severe limitations, with a high level of false-positive and false-negative results, CAD increased the examiners' sensitivity. Therefore, its supportive role in diagnostics appears to be evident.

Performance improvement in computerized detection of cerebral aneurysms by retraining classifier using feedback data collected in routine reading environment

Introduction: The performance of computer-assisted detection (CAD) software depends on the quality and quantity of the dataset used for supervised learning. To realize the continuous clinical use and performance improvement of CAD software, it is necessary to continuously collect data for supervised learning in practical use and to improve CAD software by retraining with the collected data. In this study, we investigated the performance improvement of cerebral aneurysm detection software based on retraining the classifier through a simulation-based study. Methods: We collected data for retraining during the practical use of our cerebral aneurysm detection software and retrained the classifier for false positive (FP) reduction using the collected data. The effect on improving the performance was compared by changing the number of training cases and the training algorithms. Results: The performance was improved significantly ( p < .05) by retraining using additional training cases. In contrast, there were no statistical differences in the performance upon retraining among the four training algorithms for boosting. The sensitivity at 3 FPs/case was improved from 81.5% to 89.5% by retraining with additional training cases. Conclusions: The performance of the software was effectively improved by adding training cases rather than by changing the training algorithm.

Lung cancer screening with CT: Evaluation of radiologists and different computer assisted detection software (CAD) as first and second readers for lung nodule detection at different dose levels

ObjectivesTo find the best pairing of first and second reader at highest sensitivity for detecting lung nodules with CT at various dose levels. Materials and methodsAn anthropomorphic lung phantom and artificial lung nodules were used to simulate screening CT-examination at standard dose (100mAs, 120kVp) and 8 different low dose levels, using 120, 100 and 80kVp combined with 100, 50 and 25mAs. At each dose level 40 phantoms were randomly filled with 75 solid and 25 ground glass nodules (5–12mm). Two radiologists and 3 different computer aided detection softwares (CAD) were paired to find the highest sensitivity. ResultsSensitivities at standard dose were 92%, 90%, 84%, 79% and 73% for reader 1, 2, CAD1, CAD2, CAD3, respectively. Combined sensitivity for human readers 1 and 2 improved to 97%, (p1=0.063, p2=0.016). Highest sensitivities – between 97% and 99.0% – were achieved by combining any radiologist with any CAD at any dose level. Combining any two CADs, sensitivities between 85% and 88% were significantly lower than for radiologists combined with CAD (p<0.03). ConclusionsCombination of a human observer with any of the tested CAD systems provide optimal sensitivity for lung nodule detection even at reduced dose at 25mAs/80kVp.

Multiple Sclerosis: Identification of Temporal Changes in Brain Lesions with Computer-Assisted Detection Software

Multiple sclerosis (MS) is a chronic disease with a progressing and evolving course. Serial imaging with MRI is the mainstay in monitoring and managing MS patients. In this work we demonstrate the performance of a locally developed computer-assisted detection (CAD) software used to track temporal changes in brain MS lesions. CAD tracks changes in T2-bright MS lesions between two time points on a 3D high-resolution isotropic FLAIR MR sequence of the brain acquired at 3 Tesla. The program consists of an image-processing pipeline, and displays scrollable difference maps used as an aid to the neuroradiologist for assessing lesional change. To assess the value of the software we have compared diagnostic accuracy and duration of interpretation of the CAD-assisted and routine clinical interpretations in 98 randomly chosen, paired MR examinations from 88 patients (68 women, 20 men, mean age 43.5, age range 21-75) with a diagnosis of definite MS. The ground truth was determined by a three-expert panel. In case-wise analysis, CAD interpretation showed higher sensitivity than a clinical report (87% vs 77%, respectively). Lesion-wise analysis demonstrated improved sensitivity of CAD over a routine clinical interpretation of 40%-48%. Mean software-assisted interpretation time was 2.7 min. Our study demonstrates the potential of including CAD software in the workflow of neuroradiology practice for the detection of MS lesional change. Automated quantification of temporal change in MS lesion load may also be used in clinical research, e.g., in drug trials.

Computer-aided detection for CT colonography: incremental benefit of observer training

The purpose of this study was to investigate the incremental effect of focused training on observer performance when using computer-assisted detection (CAD) software to interpret CT colonography (CTC). Six radiologists who were relatively inexperienced with CTC interpretation underwent 1 day of focused training before reading 20 patient datasets with the assistance of CAD software (ColonCAR 1.3, Medicsight PLC). Sensitivity, specificity and interpretation times were determined and compared with previous performance when reading the same datasets but without the benefit of focused training, using the binomial exact test and Wilcoxon's signed rank test. Per-polyp sensitivity improved after training by 18% overall (95% confidence interval (CI): 14-24%, p<0.001) and was greatest for polyps of 6-9 mm (26%, 95% CI: 18-34%, p<0.001). Absolute sensitivity was 23% (9-36%), 51% (33-71%) and 74% (44-100%) for polyps of <or=5 mm, 6-9 mm and >or=10 mm, respectively. Specificity fell significantly after focused training (median of 5.5 false positives per 20 datasets (interquartile range (IQR): 4-6) post-training vs median of 2.5 (IQR: 1-5) pre-training, p = 0.03). Interpretation time also increased significantly after training (from a median of 9.3 min (IQR: 9.3-14.5 min) to a median of 17.1 min (IQR: 15.4-19.4 min), p = 0.03). In conclusion, one day of training increases observer polyp sensitivity when using CAD for CTC at the expense of increased reporting time and reduction in specificity.

Computer assisted detection software for CT colonography: effect of sphericity filter on performance characteristics for patients with and without fecal tagging

The aim of this study is to investigate the effect of changing sphericity filter values on performance of a computer assisted detection (CAD) system for CT colonography for data with and without fecal tagging. Colonography data from 138 patients with 317 validated polyps were divided into those with (86) and without (52) fecal tagging. Polyp coordinates were established by three observers and datasets analysed subsequently by a proprietary CAD system used at four discrete sphericity filter settings. Prompts were compared with the known coordinates in order to determine sensitivity and specificity. Sensitivity was highest at low sphericity; of 164 polyps 6 mm or more, 144 (87.8%) were detected at sphericity 0.3, and 132 (80.1%) at sphericity 0.9. Of 42 polyps measuring 10 mm or more, 40 (95.2%) were detected at sphericity 0.3, and 36 (85.7%) at sphericity 0.9. There was no significant difference in sensitivity for tagged and un-tagged data but specificity was reduced in tagged data at low sphericity and significantly reduced in untagged data at high sphericity. CAD had a sensitivity of 95.2% for polyps measuring 1 cm or more and 87.8% for polyps 6 mm or more when used at a sphericity setting of 0.3. Higher sphericity settings increased specificity while reducing sensitivity. The bowel preparation used significantly impacts on specificity.