ABSTRACT
Purpose: Although there are several options for improving the generalizability of learned models, a data instance-based approach is desirable when stable data acquisition conditions cannot be guaranteed. Despite the wide use of data transformation methods to reduce data discrepancies between different data domains, detailed analysis for explaining the performance of data transformation methods is lacking. Approach: This study compares several data transformation methods in the tuberculosis detection task with multi-institutional chest x-ray (CXR) data. Five different data transformations, including normalization, standardization with and without lung masking, and multi-frequency-based (MFB) standardization with and without lung masking were implemented. A tuberculosis detection network was trained using a reference dataset, and the data from six other sites were used for the network performance comparison. To analyze data harmonization performance, we extracted radiomic features and calculated the Mahalanobis distance. We visualized the features with a dimensionality reduction technique. Through similar methods, deep features of the trained networks were also analyzed to examine the models' responses to the data from various sites. Results: From various numerical assessments, the MFB standardization with lung masking provided the highest network performance for the non-reference datasets. From the radiomic and deep feature analyses, the features of the multi-site CXRs after MFB with lung masking were found to be well homogenized to the reference data, whereas the others showed limited performance. Conclusions: Conventional normalization and standardization showed suboptimal performance in minimizing feature differences among various sites. Our study emphasizes the strengths of MFB standardization with lung masking in terms of network performance and feature homogenization.
ABSTRACT
OBJECTIVES: This study aimed to compare histological outcomes between pre-menopausal and post-menopausal women with cervical cytological abnormalities and to investigate the clinical factors affecting the misinterpretation of cytology and histology. METHODS: We conducted a retrospective analysis of 599 patients with abnormal cervical cytology who underwent loop electrosurgical excision procedure (LEEP) between January 2010 and May 2019. Baseline characteristics were collected, including age, height, weight, body mass index, gravity, parity, and menopausal status. In total, 477 pre-menopausal women and 122 post-menopausal women were recruited. RESULTS: Atypical squamous cells of undetermined significance (ASC-US) or low-grade squamous intraepithelial lesions were cytologically observed in 73.4% (135/184) of the pre-menopausal women, which were high-grade lesions confirmed by LEEP. In post-menopausal patients with cytology results that cannot exclude high-grade squamous intraepithelial lesions (ASC-H) or high-grade squamous intraepithelial lesions (HSIL), 27.0% (24/89) were confirmed to have histologically low-grade lesions. High-risk HPV (hrHPV) prevalence in abnormal cervical smears was 92.2%. Moreover, other hrHPVs had a higher risk of unexpected histological outcomes unrelated to cytologic results. CONCLUSIONS: Menopausal status and HPV infection are associated with misinterpretation of cervical cytology and histology. Therefore, the menopausal status of patients should be considered for the management of cervical cytology, and primary co-testing is recommended to identify women at risk of cervical abnormalities.
ABSTRACT
We report a fully automated DNA purification platform with a micropored membrane in the channel utilizing centrifugal microfluidics on a lab-on-a-disc (LOD). The microfluidic flow in the LOD, into which the reagents are injected for DNA purification, is controlled by a single motor and laser burst valve. The sample and reagents pass successively through the micropored membrane in the channel when each laser burst valve is opened. The Coriolis effect is used by rotating the LOD bi-directionally to increase the purity of the DNA, thereby preventing the mixing of the waste and elution solutions. The total process from the lysed sample injection into the LOD to obtaining the purified DNA was finished within 7 min with only one manual step. The experimental result for Salmonella shows that the proposed microfluidic platform is comparable to the existing devices in terms of the purity and yield of DNA.
