CASE PRESENTATION: A young man involved in a car accident and sustained blunt thoracic injuries, among others. As part of primary survey, FAST scan was performed. Subxiphoid view to look for evidence of pericardial effusion showed part of the cardiac image obscured by A-lines. Other cardiac windows showed only A-lines, as well. A suspicion of pneumopericardium was raised and CT scan confirmed the diagnosis.
CONCLUSIONS: Although FAST scan was originally used to look for presence of free fluid, with the knowledge of lung ultrasound for pneumothorax, our findings suggest that FAST scan can also be used to detect pneumopericardium.
MATERIALS AND METHODS: Eighteen (18) participants who met the inclusion and exclusion criteria were scanned using a standard non-contrast MRI shoulder protocol including the PDFS pulse sequence and the PROPELLER PDFS pulse sequence using a small flex coil and a dedicated shoulder coil. Two experienced musculoskeletal (MSK) radiologists evaluated and graded the presence of artifacts on the MR images and the SNR and CNR were measured quantitatively.
RESULTS: The non-parametric Wilcoxon Signed Rank test revealed a significant reduction in motion and pulsation artifacts between the PROPELLER PDFS pulse sequence and the standard PDFS pulse sequence. In addition, the nonparametric Mann-Whitney U test revealed that the mean rank of SNR for the standard sequence was statistically significant when compared to the PROPELLER sequence for both coil types. The CNR of the PROPELLER sequence was statistically significant between fat-fluid, bone-fluid, bonetendon, bone-muscle, and muscle-fluid when using SFC and DSC.
CONCLUSION: This study proved that the PROPELLER-PDFS pulse sequence effectively eliminates motion and pulsation artifacts, regardless of the coils utilised. The PROPELLERPDFS pulse sequence can therefore be implemented into the standard MRI shoulder procedure.
AIM: To compare the quality of CT brain images produced by a fixed CT scanner and a portable CT scanner (CereTom).
METHODS: This work was a single-centre retrospective study of CT brain images from 112 neurosurgical patients. Hounsfield units (HUs) of the images from CereTom were measured for air, water and bone. Three assessors independently evaluated the images from the fixed CT scanner and CereTom. Streak artefacts, visualisation of lesions and grey-white matter differentiation were evaluated at three different levels (centrum semiovale, basal ganglia and middle cerebellar peduncles). Each evaluation was scored 1 (poor), 2 (average) or 3 (good) and summed up to form an ordinal reading of 3 to 9.
RESULTS: HUs for air, water and bone from CereTom were within the recommended value by the American College of Radiology (ACR). Streak artefact evaluation scores for the fixed CT scanner was 8.54 versus 7.46 (Z = -5.67) for CereTom at the centrum semiovale, 8.38 (SD = 1.12) versus 7.32 (SD = 1.63) at the basal ganglia and 8.21 (SD = 1.30) versus 6.97 (SD = 2.77) at the middle cerebellar peduncles. Grey-white matter differentiation showed scores of 8.27 (SD = 1.04) versus 7.21 (SD = 1.41) at the centrum semiovale, 8.26 (SD = 1.07) versus 7.00 (SD = 1.47) at the basal ganglia and 8.38 (SD = 1.11) versus 6.74 (SD = 1.55) at the middle cerebellar peduncles. Visualisation of lesions showed scores of 8.86 versus 8.21 (Z = -4.24) at the centrum semiovale, 8.93 versus 8.18 (Z = -5.32) at the basal ganglia and 8.79 versus 8.06 (Z = -4.93) at the middle cerebellar peduncles. All results were significant with P-value < 0.01.
CONCLUSIONS: Results of the study showed a significant difference in image quality produced by the fixed CT scanner and CereTom, with the latter being more inferior than the former. However, HUs of the images produced by CereTom do fulfil the recommendation of the ACR.
OBJECTIVE: This paper presents a machine learning-based approach for the automatic classification of regular and irregular capnogram segments.
METHODS: Herein, we proposed four time- and two frequency-domain features experimented with the support vector machine classifier through ten-fold cross-validation. MATLAB simulation was conducted on 100 regular and 100 irregular 15 s capnogram segments. Analysis of variance was performed to investigate the significance of the proposed features. Pearson's correlation was utilized to select the relatively most substantial ones, namely variance and the area under normalized magnitude spectrum. Classification performance, using these features, was evaluated against two feature sets in which either time- or frequency-domain features only were employed.
RESULTS: Results showed a classification accuracy of 86.5%, which outperformed the other cases by an average of 5.5%. The achieved specificity, sensitivity, and precision were 84%, 89% and 86.51%, respectively. The average execution time for feature extraction and classification per segment is only 36 ms.
CONCLUSION: The proposed approach can be integrated with capnography devices for real-time capnogram-based respiratory assessment. However, further research is recommended to enhance the classification performance.