Assessment of Radiology Artificial Intelligence Software: A Validation and Evaluation Framework.

Artificial intelligence (AI) software in radiology is becoming increasingly prevalent and performance is improving rapidly with new applications for given use cases being developed continuously, oftentimes with development and validation occurring in parallel. Several guidelines have provided reporting standards for publications of AI-based research in medicine and radiology. Yet, there is an unmet need for recommendations on the assessment of AI software before adoption and after commercialization. As the radiology AI ecosystem continues to grow and mature, a formalization of system assessment and evaluation is paramount to ensure patient safety, relevance and support to clinical workflows, and optimal allocation of limited AI development and validation resources before broader implementation into clinical practice. To fulfil these needs, we provide a glossary for AI software types, use cases and roles within the clinical workflow; list healthcare needs, key performance indicators and required information about software prior to assessment; and lay out examples of software performance metrics per software category. This conceptual framework is intended to streamline communication with the AI software industry and provide healthcare decision makers and radiologists with tools to assess the potential use of these software. The proposed software evaluation framework lays the foundation for a radiologist-led prospective validation network of radiology AI software. Learning Points: The rapid expansion of AI applications in radiology requires standardization of AI software specification, classification, and evaluation. The Canadian Association of Radiologists' AI Tech & Apps Working Group Proposes an AI Specification document format and supports the implementation of a clinical expert evaluation process for Radiology AI software.

Safety and Efficacy of Endovascular Fiducial Marker Insertion for CyberKnife Stereotactic Radiation Therapy Planning in Early-Stage Lung Cancer.

PURPOSE: To evaluate the safety and efficacy of endovascular fiducial markers (EVFMs) for CyberKnife stereotactic radiation therapy (RT) of malignant lung neoplasms in patients with contraindications to percutaneous fiducial marker (PTFM) placement. MATERIALS AND METHODS: Between January 2011 and December 2013, 15 patients (7 men and 8 women) aged 59-87 years (mean, 73 y) underwent EVFM placement and 109 patients (54 men and 55 women) aged 55-92 years (mean, 73 y) underwent 114 PTFM placements. All patients who received EVFMs had contraindications to PTFM insertion. Complications were recorded for all 129 procedures. Proportions of markers used for CyberKnife stereotactic RT planning and distance between markers and tumor were assessed in all 15 patients in the EVFM group and in 15 randomly selected patients in the PTFM group. RESULTS: Sixty-two EVFMs and 56 PTFMs were inserted in the groups of 15 patients. CyberKnife stereotactic RT was not performed in 2 EVFM recipients (8 EVFMs) and 1 PTFM recipient (4 PTFMs). CyberKnife stereotactic RT was planned with the use of 39 of 54 EVFMs (72%) in 13 patients and 37 of 52 PTFMs (71%) in 14 patients. Mean distances between the tumor and EVFMs and PTFMs were 5.5 mm and 2.7 mm, respectively (P = .0152). No complications occurred in the EVFM group. In the PTFM group, 60 of 114 patients (52.6%) had small or moderate pneumothoraxes, 13 (11.4%) had chest tubes inserted, and 5 (4.4%) had self-limiting hemoptysis. CONCLUSIONS: EVFM is feasible and safe, and may therefore offer an alternative when patients are not candidates for percutaneous marker placement.

Evaluation of peripheral pulmonary arteries at 80 kV and at 140 kV: dual-energy computed tomography assessment in 51 patients.

PURPOSE: To compare peripheral pulmonary artery image quality at 80 kVp and 140 kVp in the same patients. MATERIALS AND METHODS: Image quality of third-, fourth- and fifth-order arteries was assessed at 80 kV and 140 kV on 1-mm-thick transverse scans, generated from dual-source computed tomography (CT) acquisitions. RESULTS: The mean level of enhancement was significantly higher at 80 kV compared with 140 kV for the third-, fourth-, and fifth-order arteries (P < 0.0001). Despite a higher noise level at 80 kV (P < 0.0001), the signal-to-noise ratio and contrast-to-noise ratio were significantly higher at 80 kV than at 140 kV at the level of third-, fourth-, and fifth-order arteries (P < 0.0001). The mean vascular attenuation, mean signal-to-noise ratio and contrast-to-noise ratio for peripheral arteries were significantly superior at 80 kV in the 3 body mass index categories (P < 0.005). CONCLUSIONS: Eighty-kilovolt protocols significantly improve the image quality of peripheral pulmonary arteries on CT angiograms of the chest.

Anomalous right vertebral artery arising from the arch of aorta: report of three cases.

The right vertebral artery most commonly originates as the first branch of the right subclavian artery. Although anatomical variants of the aortic arch are commonly encountered on cross-sectional imaging, certain variants of the right vertebral artery are exceedingly rare. In this report, we present 3 cases of aberrant right vertebral artery arising as the last branch of the aortic arch, a very rare variant.

Assessing the Need for Adjuvant Chemotherapy After Stereotactic Body Radiation Therapy in Early-stage Non-small Cell Lung Carcinoma.

PURPOSE: Surgery remains the standard treatment for medically operable patients with early-stage non-small cell lung carcinoma (NSCLC). Following surgical resection, adjuvant chemotherapy is recommended for large tumors >4 cm. For unfit patients, stereotactic body radiation therapy (SBRT) has emerged as an excellent alternative to surgery. This study aims to assess patterns of recurrence and discuss the role of chemotherapy after SBRT for NSCLC. METHODS: We reviewed patients treated with SBRT for primary early-stage NSCLC between 2009 and 2015. Total target doses were between 50 and 60 Gy administered in three to eight fractions. All patients had a staging fluorodeoxyglucose (FDG) positron emission tomography (PET) integrated with computed tomography (CT) scan, and histologic confirmation was obtained whenever possible. Mediastinal staging was performed if lymph node involvement was suspected on CT or PET/CT. Survival outcomes were estimated using the Kaplan-Meier method. RESULTS: Among the 559 early-stage NSCLC patients treated with SBRT, 121 patients were stage T2N0. The one-year and three-year overall survival rates were 88% and 70%, respectively, for patients with T2 disease, compared to 95% and 81%, respectively, for the T1 patients (p<0.05). The one-year and three-year local control rates were equal in both groups (98% and 91%, respectively). In T2 patients, 25 (21%) presented a relapse, among which 21 (84%) were nodal or distant. The median survival of T2N0 patients following a relapse was 11 months. CONCLUSION: Lung SBRT provides high local control rates, even for larger tumors. When patients relapse, the majority of them do so at regional or distant sites. These results raise the question as to whether adjuvant treatment should be considered following SBRT for larger tumors.

Percutaneous CT-guided lung interventions-local pleural anesthesia.

Local pleural anesthesia during percutaneous lung interventions is an important part of the procedure that has not been standardized in the literature. Significant pain can be experienced during lung biopsies, which may affect the outcome of intervention. We describe a step-by-step method of anesthesia targeting the most sensitive anatomical structure involved: the parietal pleura.

Long-term quality of life in early-stage non-small cell lung cancer patients treated with robotic stereotactic ablative radiation therapy.

PURPOSE: The purpose of this study was to prospectively evaluate the quality of life (QoL) and pulmonary function of patients with early-stage non-small cell lung cancer treated with robotic stereotactic ablative radiation therapy (SABR). METHODS AND MATERIALS: Eligible patients all had histologically confirmed stage I non-small cell lung cancer and were not surgical candidates because of poor pulmonary function, comorbidities, or refusal of surgery. SABR was delivered at a median dose of 60 Gy in 3 fractions for peripheral tumors and 50 Gy in 4 or 5 fractions for central tumors. QoL was scored using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 (QLQ-C30) and Lung Cancer-13 questionnaires. Pulmonary function tests (PFTs) included forced expiratory volume in 1 second (FEV1) and lung diffusion capacity. Changes over time in QoL scores and PFTs were tested with nonparametric tests for longitudinal data. Local control, survival, and toxicities are also presented. RESULTS: From January 2010 to May 2013, 45 patients were enrolled. Median follow-up was 41 months. QLQ-C30 mean baseline scores for global QoL and physical functioning were 66 ± 20% and 73 ± 22%. Multilevel analyses showed no statistically and clinically significant (10-point change) deterioration in any of the QoL scores after SABR. Mean baseline FEV1 was 1.39 ± 0.51 L, and mean lung diffusion capacity was 63 ± 25% of predicted. We saw no significant change in PFTs at any time point. At 3 years, local control, disease-free survival, and overall survival were, respectively, 94%, 67%, and 75%. CONCLUSIONS: In nonsurgical patients with multiple comorbidities, lung SABR achieves long-term local control while maintaining QoL and pulmonary function.

Repair of a bronchovascular fistula four years after right carinal pneumonectomy.

Bronchovascular fistula is a rare but serious complication that usually presents with massive hemoptysis and is associated with a high risk of mortality. Factors leading to the development of the fistula include altered bronchial anastomotic healing, presence of granulation tissue, devascularization of the bronchial tree, ischemia of the airway, and poor anastomotic technique. Bronchovascular fistulae usually occur in the early postoperative period and are reported in 1% to 3% of the patients undergoing a bronchoplastic procedure. We report the case of a 53-year-old woman who presented with massive hemoptysis secondary to a bronchovascular fistula 4 years after right carinal pneumonectomy.

Postoperative imaging after lung transplantation.

Lung transplantation (LT) is an established procedure for chronic end-stage lung diseases. Complications are frequent and diverse and are the consequence of the complex surgical technique, the severity of the initial pathology, and the deep state of posttransplantation immunosuppression. Complications following LT include primary graft dysfunction, rejection (hyperacute, acute, and chronic), infections, posttransplantation lymphoproliferative disease, pleural and airway complications, native lung complications, and recurrence of primary disease. An understanding of these complications, their temporal evolution, and the role of radiology and other diagnostic methods in their diagnosis and management will help reduce the morbidity and mortality associated with LT.

Nonmassive acute pulmonary embolism: evaluation of the impact of pulmonary arterial wall distensibility on the assessment of the CT obstruction score.

PURPOSE: To evaluate the impact of pulmonary arterial wall distensibility on the assessment of a computed tomography (CT) score in patients with nonmassive pulmonary embolism (PE) (ie, Mastora score). MATERIALS AND METHODS: The arterial wall distensibility of five central pulmonary arteries (pulmonary artery trunk, right and left main pulmonary arteries, right and left interlobar pulmonary arteries) was studied on ECG-gated CT angiographic studies of the chest in 15 patients with no pulmonary arterial hypertension (group 1; mean pulmonary artery pressure: 17.2 mm Hg) and 9 patients with nonmassive PE (group 2), using 2D reconstructions at every 10% of the R-R interval. RESULTS: The systolic and diastolic reconstruction time windows of the examined arteries were identical in the 2 groups, obtained at 20% and 80% of the R-R interval, respectively. No statistically significant difference was observed between the mean values of the pulmonary arterial wall distensibility between the 2 groups, varying between 20.5% and 24% in group 1 and between 23.3% and 25.9% in group 2. The coefficients of variation of the average arterial surfaces were found to vary between 4.30% and 6.50% in group 1 and 4.2% and 8.4% in group 2. Except the pulmonary artery trunk in group 2, all the intraclass correlation coefficients were around 0.8 or greater than 0.8, that is the cutoff for good homogeneity of measurements. CONCLUSION: The pulmonary arterial wall systolic-diastolic distensibility does not interfere with the assessment of a CT obstruction score in the setting of nonmassive PE.

Thoracic applications of dual-source CT technology.

Among the various imaging modalities available, CT has remained over time the core imaging technique for the evaluation of respiratory disorders. The recent advent of dual-source CT offers innovative approaches to investigate thoracic diseases, based on the use of one or two tubes as well as single or dual energy to scan the entire thorax. Two major options can be used in clinical practice with promising results. Dual source, single-energy scanning allows scanning of the entire thorax with ultra-high temporal resolution which opens the field of integrated cardiothoracic imaging without ECG gating as well as optimized evaluation of pediatric and adult patients with limited ability to cooperate. Dual-source, dual-energy acquisitions represent another very innovative means of investigating respiratory disorders, adding tissue characterization and functional analysis to morphological evaluation. The purpose of this review article is to provide results on preliminary experiences with the above-mentioned scanning conditions with dual-source CT and to envisage potential forthcoming applications in the field of thoracic imaging.