Determination of disease severity in COVID-19 patients using deep learning in chest X-ray images.

PURPOSE: Chest X-ray plays a key role in diagnosis and management of COVID-19 patients and imaging features associated with clinical elements may assist with the development or validation of automated image analysis tools. We aimed to identify associations between clinical and radiographic features as well as to assess the feasibility of deep learning applied to chest X-rays in the setting of an acute COVID-19 outbreak. METHODS: A retrospective study of X-rays, clinical, and laboratory data was performed from 48 SARS-CoV-2 RT-PCR positive patients (age 60±17 years, 15 women) between February 22 and March 6, 2020 from a tertiary care hospital in Milan, Italy. Sixty-five chest X-rays were reviewed by two radiologists for alveolar and interstitial opacities and classified by severity on a scale from 0 to 3. Clinical factors (age, symptoms, comorbidities) were investigated for association with opacity severity and also with placement of central line or endotracheal tube. Deep learning models were then trained for two tasks: lung segmentation and opacity detection. Imaging characteristics were compared to clinical datapoints using the unpaired student's t-test or Mann-Whitney U test. Cohen's kappa analysis was used to evaluate the concordance of deep learning to conventional radiologist interpretation. RESULTS: Fifty-six percent of patients presented with alveolar opacities, 73% had interstitial opacities, and 23% had normal X-rays. The presence of alveolar or interstitial opacities was statistically correlated with age (P = 0.008) and comorbidities (P = 0.005). The extent of alveolar or interstitial opacities on baseline X-ray was significantly associated with the presence of endotracheal tube (P = 0.0008 and P = 0.049) or central line (P = 0.003 and P = 0.007). In comparison to human interpretation, the deep learning model achieved a kappa concordance of 0.51 for alveolar opacities and 0.71 for interstitial opacities. CONCLUSION: Chest X-ray analysis in an acute COVID-19 outbreak showed that the severity of opacities was associated with advanced age, comorbidities, as well as acuity of care. Artificial intelligence tools based upon deep learning of COVID-19 chest X-rays are feasible in the acute outbreak setting.

Boiling histotripsy lesion characterization on a clinical magnetic resonance imaging-guided high intensity focused ultrasound system.

PURPOSE: High intensity focused ultrasound (HIFU) is a non-invasive therapeutic technique that can thermally ablate tumors. Boiling histotripsy (BH) is a HIFU approach that can emulsify tissue in a few milliseconds. Lesion volume and temperature effects for different BH sonication parameters are currently not well characterized. In this work, lesion volume, temperature distribution, and area of lethal thermal dose were characterized for varying BH sonication parameters in tissue-mimicking phantoms (TMP) and demonstrated in ex vivo tissues. METHODS: The following BH sonication parameters were varied using a clinical MR-HIFU system (Sonalleve V2, Philips, Vantaa, Finland): acoustic power, number of cycles/pulse, total sonication time, and pulse repetition frequency (PRF). A 3×3×3 pattern was sonicated inside TMP's and ex vivo tissues. Post sonication, lesion volumes were quantified using 3D ultrasonography and temperature and thermal dose distributions were analyzed offline. Ex vivo tissues were sectioned and stained with H&E post sonication to assess tissue damage. RESULTS: Significant increase in lesion volume was observed while increasing the number of cycles/pulse and PRF. Other sonication parameters had no significant effect on lesion volume. Temperature full width at half maximum at the end of sonication increased significantly with all parameters except total sonication time. Positive correlation was also found between lethal thermal dose and lesion volume for all parameters except number of cycles/pulse. Gross pathology of ex vivo tissues post sonication displayed either completely or partially damaged tissue at the focal region. Surrounding tissues presented sharp boundaries, with little or no structural damage to adjacent critical structures such as bile duct and nerves. CONCLUSION: Our characterization of effects of HIFU sonication parameters on the resulting lesion demonstrates the ability to control lesion morphologic and thermal characteristics with a clinical MR-HIFU system in TMP's and ex vivo tissues. We demonstrate that this system can produce spatially precise lesions in both phantoms and ex vivo tissues. The results provide guidance on a preliminary set of BH sonication parameters for this system, with a potential to facilitate BH translation to the clinic.