Artificial intelligence: radiologists' expectations and opinions gleaned from a nationwide online survey.

PURPOSE: To report the results of a nationwide online survey on artificial intelligence (AI) among radiologist members of the Italian Society of Medical and Interventional Radiology (SIRM). METHODS AND MATERIALS: All members were invited to the survey as an initiative by the Imaging Informatics Chapter of SIRM. The survey consisted of 13 questions about the participants' demographic information, perceived advantages and issues related to AI implementation in radiological practice, and their overall opinion about AI. RESULTS: In total, 1032 radiologists (equaling 9.5% of active SIRM members for the year 2019) joined the survey. Perceived AI advantages included a lower diagnostic error rate (750/1027, 73.0%) and optimization of radiologists' work (697/1027, 67.9%). The risk of a poorer professional reputation of radiologists compared with non-radiologists (617/1024, 60.3%), and increased costs and workload due to AI system maintenance and data analysis (399/1024, 39.0%) were seen as potential issues. Most radiologists stated that specific policies should regulate the use of AI (933/1032, 90.4%) and were not afraid of losing their job due to it (917/1032, 88.9%). Overall, 77.0% of respondents (794/1032) were favorable to the adoption of AI, whereas 18.0% (186/1032) were uncertain and 5.0% (52/1032) were unfavorable. CONCLUSIONS: Radiologists had a mostly positive attitude toward the implementation of AI in their working practice. They were not concerned that AI will replace them, but rather that it might diminish their professional reputation.

Impact of coronavirus disease 2019 (COVID-19) emergency on Italian radiologists: a national survey.

OBJECTIVES: To perform an online survey aimed at evaluating the impact of COVID-19 on Italian radiology departments. METHODS: We launched a survey composed of 25 questions about how COVID-19 has changed the safety and organization of daily activity in Italian radiology units. RESULTS: A total of 2136/10,564 (20.2%) radiologists of the Italian Society of Medical and Interventional Radiology participated. Two-thirds performed at least one diagnostic/interventional procedure on COVID-19 patients. The 88.1% reported a reduction in the elective imaging volumes, with US, mammography, and MRI having shown the greater decrease (41.1%, 23.9%, and 21.1%, respectively). In 69.6% of cases, institutions had trouble getting personal protective equipment (PPE), especially public hospitals and southern institutions. Less than 30% of participants were subjected to RT-PCR swab test, although 81.5% believed that it should be done on all health workers and 70% suggested it as the most important measure to improve safety at work. Slightly more than half of participants declared to work safely and felt to be adequately protected by their institutions. Up to 20% of northern participants were redeployed to clinical services. The first imaging examination performed by admitted COVID-19 patients was chest radiography in 76.3% of cases. Almost half of participants reported that less than 30% of health workers were infected in their radiology department, with higher rates in northern regions and public institutions. CONCLUSIONS: This snapshot of the current situation in Italian radiology departments could be used to harmonize the organization of working activity in order to safely and effectively face this pandemic. KEY POINTS: • More than two-thirds of institutions had trouble getting PPE for health workers, with public hospitals and southern institutions that presented more procurement problems • A substantial drop of imaging volumes was observed in the vast majority of Italian radiology departments, mostly due to the decrease of ultrasound, mammography, and MRI, especially in private practice were working activity was stopped in 13.3% of institutions • RT-PCR swab to health workers was reported as the most suggested measure by Italian radiologists to improve safety at work, as more than 80% of them believed that it should be performed to all health workers, although less than 30% were subjected to this test.

Artificial intelligence: Who is responsible for the diagnosis?

The aim of the paper is to find an answer to the question "Who or what is responsible for the benefits and harms of using artificial intelligence in radiology?" When human beings make decisions, the action itself is normally connected with a direct responsibility by the agent who generated the action. You have an effect on others, and therefore, you are responsible for what you do and what you decide to do. But if you do not do this yourself, but an artificial intelligence system, it becomes difficult and important to be able to ascribe responsibility when something goes wrong. The manuscript addresses the following statements: (1) using AI, the radiologist is responsible for the diagnosis; (2) radiologists must be trained on the use of AI since they are responsible for the actions of machines; (3) radiologists involved in R&D have the responsibility to guide the respect of rules for a trustworthy AI; (4) radiologist responsibility is at risk of validating the unknown (black box); (5) radiologist decision may be biased by the AI automation; (6)risk of a paradox: increasing AI tools to compensate the lack of radiologists; (7) need of informed consent and quality measures. Future legislation must outline the contours of the professional's responsibility, with respect to the provision of the service performed autonomously by AI, balancing the professional's ability to influence and therefore correct the machine, limiting the sphere of autonomy that instead technological evolution would like to recognize to robots.

Dematerialisation of patient's informed consent in radiology: insights on current status and radiologists' opinion from an Italian online survey.

PURPOSE: To assess the current status of patient's informed consent (PIC) management at radiological centres and the overall opinion of radiologist active members of the Italian Society of Medical Radiology (SIRM) about PIC dematerialisation through an online survey. METHODS AND MATERIALS: All members were invited to join the survey as an initiative by the Imaging Informatics Chapter of SIRM. The survey consisted of 11 multiple-choice questions about participants' demographics, current local modalities of PIC acquisition and storage, perceived advantages and disadvantages of PIC dematerialisation over conventional paper-based PIC, and overall opinion about PIC dematerialisation. RESULTS: A total of 1791 radiologists (amounting to 17.4% of active SIRM members for the year 2016) joined the survey. Perceived advantages of PIC dematerialisation were easier and faster PIC recovery (96.5%), safer storage and conservation (94.5%), and reduced costs (90.7%). Conversely, the need to create dedicated areas for PIC acquisition inside each radiological unit (64.0%) and to gain preliminary approval for the use of advanced digital signature tools from patients (51.8%) were seen as potential disadvantages. Overall, 94.5% of respondents had a positive opinion about PIC dematerialisation. CONCLUSION: Radiologists were mostly favourable to PIC dematerialisation. However, concerns were raised that its practical implementation might face hurdles due to its complexity in current real life working conditions.

Results of an Italian survey on teleradiology.

OBJECTIVES: The aim of this study is to present the results of the Italian survey on teleradiology (TR). METHODS: Two radiologists created an online electronic survey using the Survey Monkey web-based tool. The questionnaire was then improved by suggestions from a multidisciplinary group of experts. In its final form, the survey consisted of 19 multiple-choice questions. Space was left below each question for participants to add their personal comments. Members of Italian Society of Medical Radiology (SIRM) were given 2 weeks to perform the survey. RESULTS: A total of 1599 radiologists, corresponding to 17 % of all SIRM radiologists, participated into the online survey. As a result, 62 % of participants have a positive opinion on teleradiology, while 80 % including 18 % with a negative opinion believe that teleradiology will have a future. 55 % of responders (n = 874) use teleradiology in their clinical practice. The majority of users adopt intra-mural teleradiology for coverage of emergencies (47 %), of night and weekend shifts (37 %) or to even out distribution workload (33 %). Most responders still show concern on the use of teleradiology. In particular, they think that teleradiology is too impersonal (40 %), and that it is responsible for insufficient communication with the referring clinician (39 %). CONCLUSIONS: The majority of Italian radiologists are favorable to teleradiology. However, they have concerns that teleradiology may further reduce communication with the referring clinician ad patient.

Detection of Pulmonary tuberculosis: comparing MR imaging with HRCT.

BACKGROUND: Computer Tomography (CT) is considered the gold standard for assessing the morphological changes of lung parenchyma. Although novel CT techniques have substantially decreased the radiation dose, radiation exposure is still high. Magnetic Resonance Imaging (MRI) has been established as a radiation- free alternative to CT for several lung diseases, but its role in infectious diseases still needs to be explored further. Therefore, the purpose of our study was to compare MRI with high resolution CT (HRCT) for assessing pulmonary tuberculosis. METHODS: 50 patients with culture-proven pulmonary tuberculosis underwent chest HRCT as the standard of reference and were evaluated by MRI within 24 h after HRCT. Altogether we performed 60 CT and MRI examinations, because 10 patients were also examined by CT and MRI at follow- up. Pulmonary abnormalities, their characteristics, location and distribution were analyzed by two readers who were blinded to the HRCT results. RESULTS: Artifacts did not interfere with the diagnostic value of MRI. Both HRCT and MRI correctly diagnosed pulmonary tuberculosis and identified pulmonary abnormalities in all patients. There were no significant differences between the two techniques in terms of identifying the location and distribution of the lung lesions, though the higher resolution of MRI did allow for better identification of parenchymal dishomogeneity, caseosis, and pleural or nodal involvement. CONCLUSION: Technical developments and the refinement of pulse sequences have improved the quality and speed of MRI. Our data indicate that in terms of identifying lung lesions in non-AIDS patients with non- miliary pulmonary tuberculosis, MRI achieves diagnostic performances comparable to those obtained by HRCT but with better and more rapid identification of pulmonary tissue abnormalities due to the excellent contrast resolution.

Sonographic assessment of facial HIV-related lypoatrophy.

OBJECTIVE: To investigate the utility of ultrasonography (US) for assessing and grading facial lypoatrophy (FLA) in patients with HIV. DESIGN: The social effect of FLA is huge and may reduce antiretroviral therapy adherence. Strategies for the early detection of FLA are crucial, because complete correction of FLA in late stages is unlikely. METHODS: Fifty-two HIV-positive patients undergoing highly active antiretroviral therapy underwent US with nasogenian transversal scan using a high-frequency broadband transducer (5-17 MHz) to detect FLA. Intra- and interobserver variability were calculated to assess US reproducibility. Concerning FLA grading, patients were categorized in five clinical classes and four US classes. RESULTS: Our results regarding inter- and intraobserver coefficients of variation permit the validation of US as a reproducible technique (p<.001), and a high correlation between US and clinical classification was obtained, with complete concordance for more advanced FLA classes. CONCLUSIONS: The lack of a reference objective method to quantify subcutaneous fat is a major difficulty in measuring HIV-related FLA. Our results, in accordance with data from the literature, suggest that US is an ideal tool for assessing and grading FLA. Furthermore, US may be suitable for routine evaluation in HIV-infected patients for early detection of FLA and to select its optimal management.

HIV-related pneumococcal lung disease: does highly active antiretroviral therapy or bacteremia modify radiologic appearance?

We reviewed chest radiographs of 57 HIV-infected patients with pulmonary diseases in whom Streptococcus pneumoniae was the sole respiratory pathogen isolated to evaluate whether highly active antiretroviral therapy (HAART) or bacteremia modify radiographic appearance. Pneumococcal lung disease presented as lobar pneumonia in 40% of the cases, 54% of whom were on HAART; as bronchopneumonia in 42%, 58% on HAART; as interstitial infiltrates in 17%, 60% on HAART. Bacteremia was observed 38 times in 23 patients with CD4 less than 200/mm(3), and in 15 with CD4 greater than 200/mm(3) (p > 0.05). HAART does not significantly influences radiographic appearances of lung disease caused by Streptococcus pneumoniae (p > 0.05). Immunosuppression induced by HIV infection was a major risk factor for development of pneumococcal lung disease (p = 0.04) and influences radiographic appearance; bronchopneumonia (p = 0.006), in particular multifocal (p = 0.008), which was more frequent in subjects with CD4 less than 200/mm(3). Bacteremia influences radiographic appearance of pneumococcal lung disease; lobar pneumonia was more frequent (p = 0.003), and considering CD4 cell count, was more frequent if CD4 cell count was above 200/mm(3). An original finding of this study was the frequency of interstitial changes. This pattern of pneumonia, found in 17% of our patients, could represent a difference between HIV-seropositive and -seronegative subject in displaying pneumococcal lung disease.

Progressive multifocal leukoencephalopathy in HIV-infected patients in the era of HAART: radiological features at diagnosis and follow-up and correlation with clinical variables.

The magnetic resonance (MR) imaging patterns of HIV-infected patients affected by progressive multifocal leukoencephalopathy (PML) in the HAART era have not been extensively documented. The aim of the present study is to describe the MR features of PML in HIV-infected patients at diagnosis, and the evolution during follow-up, evaluating the impact of HAART on imaging, and to correlate the MR pattern with the virological and immunological levels and with survival. We retrospectively reviewed MR imaging at baseline and at the last available follow-up within 6 months of diagnosis (median 4 months, range 1-6) of 31 HIV-positive patients affected by PML. A closer follow-up [median interval from diagnosis 39 days (range 20-139)] was also reported. At the onset of neurological disorder, 19 patients were naive for antiretroviral therapy, 7 patients were on HAART, and 5 patients were experienced but were not taking antiretroviral therapy. Upon PML diagnosis no significant differences at imaging were observed between naive and experienced patients and HAART-treated or non-HAART-treated patients. MR findings were not related to immunological status, either at baseline or at follow-up. A radiological improvement within 6 months was associated with a higher probability of a more favorable clinical evolution [OR 14.0 (2.2-87.2), p = 0.003]. The overall probability of survival at 6 months was 61.5%. A better survival was observed in patients with stable or improved MR imaging findings within 6 months [HR 4.55 (95%CI 1.36-15.19, p = 0.009]. Although HAART prolonged the survival of HIV-positive patients affected by PML, it did not seem to influence the PML MR pattern of presentation and the imaging evolution. Only the radiological outcome was predictive of clinical outcome.

Reduced computed tomography radiation dose in HIV-related pneumonia: effect on diagnostic image quality.

PURPOSE: The objective of this study was to compare the diagnostic quality of low-dose computed tomography (CT) with that of standard-dose chest CT in the diagnosis of infectious lung diseases. MATERIALS AND METHODS: Thirty chest CT scans [high-resolution computed tomography (HRCT), 15; spiral CT, 15] were performed in HIV-positive patients with an infectious lung disease. Two additional slices were obtained at two lower exposures (HRCT, 120 kV/70 mAs and 120 kV/50 mAs; spiral CT, 100 kV/56 mAs and 100 kV/40 mAs) after chest routine CT. Observers compared the quality of the images obtained at different parameters and image noise. Objective evaluation of image noise was also made. RESULTS: Diagnostic image quality was excellent in 93% of the low-dose HRCT scans and in 86% of the low-dose spiral CT scans, rates that are always acceptable in any case. Significant differences were found in noise levels between the low-dose and reference scans; however, artifacts did not compromise detection of abnormalities. In HRCT, a mean reduction of 77% from the standard technique to the low-dose scan is possible in total and in lung effective doses. In spiral CT, this reduction is lower at 71%. These values can reach a further reduction with ultra-low-dose imaging (84% in HRCT and 80% in spiral CT). CONCLUSIONS: Chest CT image quality appears to be adequate to evaluate pulmonary infectious diseases, even with an effective reduction in radiation dose. Standard-dose CT with a higher patient effective dose may be appropriate for selected cases.

Assessment of risk in radiology using malpractice RVU.

INTRODUCTION: Analysis on the causes and remedies needed to reduce the incidence of malpractice has been under continual studies, although limited data is available regarding quantitative evaluation of the risk. OBJECTIVES: To determine radiological risk in a preventive and quantitative manner and verify if the malpractice relative value units (MP-RVU) are a good indicator of associated risk factors. MATERIALS AND METHODS: Radiological examinations executed by our Radiology Department in 2000-2004 have been codified according to nomenclature HCPCS (Healthcare Common Procedure Coding System) used by United States of America Centers for Medicare and Medicaid Services (CMS). For every examination was calculated the annual weight of malpractice. The data has been groupped in macroaggregates by methodology. The ratio MP-RVU/no. examinations has been considered as an index of insurance risk (MP index) RESULTS: A total of 133,005 examinations were performed, which produced 25,252 MP-RVU points, the total mp index was 0.193. Traditional radiology represents 38% of the examinations, accounting for 8% of MP-RVU with a MP index=0.039. Ultrasound represents 35% of the examinations, accounting for 23% of MP-RVU with a MP index=0.125. CT represents 13% of the examinations, accounting for 28% of MP-RVU with a MP index=0.434. MR represents 11% of the examinations, accounting for 39% of MP-RVU with a MP index=0.667. CONCLUSIONS: Malpractice relative value units (MP-RVU) are indicative of the risk considered globally and when subgrouped. MP index correlates this risk with number of exams carried out divided by methodology. This model providing quantitative data for projects concerning risk management and in allowing the correlation between data obtained in different departments.

Comparison between indicators for the measurement of radiology activity volumes (Italy-USA).

PURPOSE: Different evaluation systems and indicators have recently been used to measure the activity volumes of Italian hospital departments, and in particular of Diagnostic Imaging Units. These measurements have mostly been based on more or less complex and repeatable indicators such as total accesses, accesses per imaging modality, type and number of exams. The aim of this study was to compare four models for measuring and evaluating productivity to assess their features and propose a common method for measuring activity volumes in a Diagnostic Imaging Unit. The models considered are: a) the numerical count, b) the model proposed by SNR-SAGO-SIRM, c) the model based on transfer prices in use in the Emilia Romagna Region (RER), d) the model used by the U.S. Health Care Financing Agency (HCFA-USA), based on a complex system of weights named RVUs (Relative Value Units). MATERIALS AND METHODS: The period under review considers two years of activity (2000-2001) at our Diagnostic Imaging Unit. The data were collected by grouping the radiological procedures into homogeneous groups (macroaggregates) which were then assessed with the four models. The reference parameters considered in order to produce homogeneous data were: the number of procedures per physician hour, the score per hour according to the SNR-SAGO-SIRM model, the score per hour according to the RER model, the number of work-RVUs per hour worked. With regard to the HCFA-USA system, the following indicators were used: the work component (work-RVU), the insurance component (malpractice RVU) and the technical component (practice expense-RVU), the equivalent units of physician time (FTE: Full Time Equivalent), such as the number of procedures per FTE, the difficulty index, and the number of RVUs per FTE. RESULTS: a) The total number of procedures was 55,884, the number of procedures per hour ranged from 2.43 (August 2000) to 4.20 (March 2000); based on the numerical count conventional radiology accounted for the most of the Unit's activity (40%). b) The total score according to the SNR-SAGO-SIRM model was 147,358; the weight of each physician hour ranged from 6.37 (August 2000) to 9.80 (October 2001). The SNR-SAGO-SIRM model indicates that the most significant macroaggregate in the Unit's activity was ultrasound (42%). c) The total score according to the RER model was 4,313,047, the weight of each physician hour varied between 159 (August 2000) and 316 (April 2000). Based on the RER model, CT (42%) accounted for most of the Unit's activity. d) According to the RVU model, the total number of work-RVUs was 37,619, and the physician weight per hour ranged from 1.45 (August 2000) to 2.86 (March 2000). The predominant method was ultrasound (35%); the number of total practice expense-RVUs was 192,749; the month with the highest score was March 2000 (9,398), while the one with the lowest score was August 2000 (4,710); the total number of malpractice RVUs was 9,940, and the months with the highest scores were April 2000 (487) and March 2000 (487), while the month with the lowest score was August 2000 (243), and the modality carrying the highest insurance risks was MRI (38%). We also calculated the number of procedures per FTE (6,141), the number of work-RVUs per FTE (4,134); the difficulty index resulting from the ratio between work-RVUs and number of procedures (0.67); the number of work-RVUs per hour worked (3.06). CONCLUSIONS: Based on the numerical count, conventional radiology and ultrasound play a predominant role (40% and 34%, respectively, total 74%). This approach therefore fails to reflect the weight of more technologically advanced procedures. The SNR-SAGO-SIRM model gives adequate importance to the combination ''number- weight of patients'' among the macroaggregates analysed. The RER model rewards the use of more expensive technologies, as it assesses the overall weight of the service and not only the weight of the radiologist's activity. The RVU model, with its distribution of weights, differentiates the different work, cost, and insurance components of the macroaggregates. It also introduces an important aspect that is new to our professional and scientific culture: evaluation of the ''insurance component'', whose role will become increasingly important in Italy. The difficulty index (work-RVUs/no. of procedures), which expresses the ratio between the number of modalities and their complexity, is particularly interesting. This index, adjusted to reflect the Italian situation, might help to assess the true technological and scientific content of the department's activity.

Cavitary pulmonary tuberculosis HIV-related.

INTRODUCTION: It was usually assumed that pulmonary tuberculosis (TB) in HIV-seropositive patients represents reactivation TB, despite the radiographic appearance frequently consistent rather with a recent disease. Hence, these radiographic features were considered "atypical". We have hypothesised that the so called "atypical" radiographic features could be due to a greater proportion of primary TB among these patients, representing the typical appearance of primary radiological pattern. MATERIAL AND METHODS: We reviewed chest imaging of 219 HIV+ patients with microbiological proven pulmonary tuberculosis, who were assessed for the presence, number, distribution of cavitations and for associated pulmonary parenchymal abnormalities, adenopathies and pleural effusion, and were classified as a primary or post-primary pattern. RESULTS: The patients with post-primary pattern were 50%, and the rate of cavitation was 63%, not wandering off the general population. Cavities still occurred with similar proportion in groups with CD4 <200 or >200cells/mm(3). CONCLUSION: We suggest that HIV-related pulmonary tuberculosis is typical in its radiological appearances, consistent with those of the general population, and this could be confirmed by the most recent molecular epidemiological techniques that allow to definitely classify the tuberculosis episodes as either primary or post-primary disease.

Reduced value of thallium-201 single-photon emission computed tomography in the management of HIV-related focal brain lesions in the era of highly active antiretroviral therapy.

To evaluate the diagnostic value of thallium-201 single-photon emission computed tomography (201Tl SPECT) in the management of focal brain disorders in the era of highly active antiretroviral therapy (HAART), a validation study of diagnostic procedure was performed in a tertiary clinical care center in Italy. Thirty-eight consecutive HIV-infected patients with neurological impairment and focal brain lesions (FBL) were enrolled in a prospective evaluation and underwent diagnostic procedures according to a standardized protocol based on modified previously released guidelines. Six out of seven PCNSL presented high uptake at 201Tl SPECT [sensitivity 86% (95% CI 42-99); specificity 77% (95% CI 58-90); positive predictive value (PPV) 46% (95% CI 20-74); negative predictive value (NPV) 96% (95% CI 78-100)]. Among toxoplasmic encephalitis (TE) cases 14 showed no uptake and 5 showed an increased uptake [sensitivity 74% (95% CI 49-90); specificity 42% (95% CI 21-66); PPV 56% (95% CI 35-75); NPV 61% (95% CI 32-85)]. Patients taking HAART were more likely to display an increased uptake of 201Tl in the cerebral lesions than patients without HAART (OR 5.07; 95% CI 1.19-21.5). Considering only the patients with diagnosis of TE, 60% of patients who showed high radionuclide uptake were taking HAART, while 79% of patients without relevant uptake were not taking HAART. As a result of the impact of HAART, the diagnostic value of 201Tl SPECT in the management of HIV-associated FBL could be substantially reduced. This observation suggests that in HAART-treated patients, this diagnostic tool be used only when combined with other more specific diagnostic markers.

Gynecomastia in male HIV patients MRI and US findings.

PURPOSE: To evaluate the role of MRI and sonography in differential diagnosis between gynecomastia and lipomastia in adult male HIV patients treated with high-activity antiretroviral therapy (HAART) for guide management. MATERIALS AND METHODS: Nineteen adult male HIV patients with enlargement of the breast, which developed during treatment with HAART, were examined with ultrasound (US) and magnetic resonance imaging (MRI). RESULTS: MRI was performed with SPIR T2WI, depicted in 95% of the patients, true gynecomastia and in the remaining 5% lipomastia; US demonstrated in all patients, in the subareolar region, a hypoechoic area, but was enable to make a differential diagnosis between glandular tissue and fat accumulation. CONCLUSION: US and MRI scan should be advised in HIV-infected patient in antiretroviral therapy with enlargement of the breast, in order to assess the best choice in managing this clinical condition. MRI with fat saturation sequences clarifies the tissue distribution in the mammary gland and helps to assess the amount of fat accumulated in the breast.

STIR sequence in infectious sacroiliitis in three patients.

Magnetic resonance imaging (MRI) provides the most detailed evaluation of the sacroiliac (SI) joint and surrounding soft tissue. Therefore, this technique represents the most sensitive and specific method in early diagnosis of infectious sacroiliitis. Among three patients diagnosed as having infectious sacroiliitis in our hospital, the short tau inversion recovery sequence (STIR) was found to be more effective than the T1 contrast-enhanced sequence, particularly in delineating all findings of the SI joint inflammation and allowing for the early detection of septic sacroiliitis.