COVID-19 disease, caused by the SARS-CoV-2 virus, was identified in December 2019 in China and declared a global pandemic in all over the world. Since then, many investigators have trying to determine its' epidemiology, etiology,pathophysiology, radiological manifestations, therapeutic strategies and outcome prediction, etc. In addition, a few academic societies as well as academic institution had proposed some category classifications for assessing COVID-19 pneumonia,and many clinicians reported their utility in routine clinical practice. Moreover, artificial Intelligence (AI) is a potentially powerful tool in the fight against the COVID-19 pandemic, when considering the above-mentioned situation in Chest Radiology. For present purposes, AI can currently be tried to demonstrate its' clinical utility for diagnosis, disease severity evaluation and treatment response assessment for radiological examination in patients with COVID-19 infection. Then,many papers have been reported radiological features as well as clinical features of COVID-19 infection, development, and clinical application of AI system for chest radiograph (or chest X-ray : CXR) or computed tomography (CT) in not only China, but also other Western countries including Japan and Korea. Encouraging for developments and clinical studies for AI may contribute to not only clinicians, but also radiologists for improving clinical practice in patients with COVID-19 infection. Moreover, AI is required a careful balance between data privacy and public health concerns, and more rigorous human-AI interaction. So, the progress of AI is very important for future control of COVID-19 infection in routine clinical practice.
In this article, we discuss 1) radiological manifestation of COVID-19 pneumonia on CXR or CT, 2) current clinical indication for CT examination in suspected COVID-19 patients, 3) previously published categorization for probability of COVID-19 pneumonia on CT (or CXR) by a few academic societies or institutions and 4) current status and future direction of AI for radiological assessment in patients with COVID-19.
To reconstruct three-energy tomograms simultaneously, we constructed a triple-sensitivity X-ray computed tomography (TS-CT) scanner to observe gradual variations in image contrast. The TS-CT scanner is a first-generation type using a room-temperature cadmium telluride (CdTe) detector with one analog and two digital amplifiers. Using a line beam,CT is performed by repeating the translation using the CdTe detector and rotation using a turntable. The CdTe detector,with a specific gravity of 5.85, is highly sensitive to X rays and detects photons penetrating an object using a 0.25-mm diameter lead pinhole for improving the spatial resolution. Without photon counting, the currents flowing through the CdTe diode are converted into voltages using a current-to-voltage amplifier and amplified by a voltage-to-voltage amplifier. The effective photon energy increased with increasing digital amplification factor at a constant maximum output of 5.06 V owing to the beam hardening by the object. In the TS-CT, the tube voltage and current were regulated as 100 kV and 1.0 mA, respectively. The spatial resolutions and scanning time for CT were approximately 0.25×0.25 mm2 and 19.6 min,respectively. Using gadolinium contrast media, the image contrast of the media varied substantially according to changes in the digital amplification factor, and blood vessels were observed at a high contrast. The digital amplifier could be used instead of a real analog amplifier at low amplification factors below 3.0.
Survival analysis is often used in radiomic studies for predicting prognosis and recurrence. A typical model for the survival analysis is Cox regression. An essential assumption when using the Cox regression model is proportional hazards. However, the proportional hazards assumption has not been verified in a lot of radiomic studies so far. In this study, we investigated the existence ratio of radiomic features that did not satisfy the proportional hazards in breast cancer and glioblastoma. Additionally, we proposed a hypothesis test of proportional hazards and an evaluation method using a scatter plot. The experimental results showed that radiomics features did not satisfy the proportional hazards assumption in 6.5% (24/369) of breast cancer and 21.4% (79/369) of glioblastoma. When the phenotype of tumor changed in a short period of time, it was found that radiomic features often did not satisfy the proportional hazards assumption. Therefore, in such radiomic studies, it is necessary to test the hypothesis of proportional hazards.