Publications by authors named "Kai-Lin Chu"
Automated Quantification of HER2 Amplification Levels Using Deep Learning.
IEEE J Biomed Health Inform
October 2024
Article Synopsis
- HER2 assessment is crucial for selecting patients for anti-HER2 treatments, but manually analyzing HER2 amplification is time-consuming and prone to errors due to subjective interpretations and complex cell imagery.
- To overcome these challenges, a new deep learning model has been developed that can accurately quantify HER2 amplification status in breast cancer by analyzing FISH and DISH datasets, achieving high accuracy and outperforming existing methods.
- The model was also successfully applied to assess HER2 amplification in gastric cancer patients, yielding promising results with high accuracy and precision, indicating its potential beyond breast cancer applications.
Dark-field (DF) optical microscopy, combined with optical simulation based on modal diffraction theory for transverse electric polarized white light, is shown to provide non-invasive, sub-wavelength geometrical information for nanoscale etched device structures. Room temperature (RT) single electron transistors (SETs) in silicon, defined using etched ∼10 nm point-contacts (PCs) and in-plane side gates, are investigated to enable fabrication fault detection. Devices are inspected using scanning electron microscopy, bright-field (BF) and DF imaging.
View Article and Find Full Text PDFArticle Synopsis
- * Traditional microscopic examination for determining HER2 gene amplification is inconsistent and time-consuming due to high variability among pathologists, affecting diagnosis accuracy.
- * This paper presents an effective deep learning method that significantly improves the detection of breast cancer and HER2 amplification, outperforming existing methods while also being faster and requiring less computational resources.
Article Synopsis
- According to the World Health Organization's 2022 report, cancer is the leading cause of death globally, responsible for nearly one in six fatalities, highlighting the need for early detection to lower mortality rates.
- The study introduces a soft label fully convolutional network (SL-FCN) designed to enhance breast cancer therapy and diagnose thyroid cancer by automatically segmenting critical features in medical images.
- Evaluations against thirteen other deep learning models show that SL-FCN demonstrates strong performance in accuracy and recall, achieving up to 94.64% accuracy in detecting HER2 amplification in breast cancer datasets and effectively segmenting papillary thyroid carcinoma in thyroid cases.
Background/purpose: Neonatal screening using tandem mass spectrometry (MS/MS) started in Taiwan in 2000. We evaluated the efficacy of this system by analyzing the frequency of diseases and the outcome of the patients identified.
Methods: Between August 2001 and July 2004, 199, 922 neonates were screened for 10 amino acids and acylcarnitines using MS/MS in a single center.
Carbohydrate deficient glycoprotein syndromes (CDG) are inherited multisystem disorders characterized by the abnormal glycosylation of a number of serum glycoproteins. CDG-Ia results from deficiency of phosphomannomutase that catalyzes the conversion of mannose-6-phosphate to mannose-1-phosphate in the cytosol. We report a case of CDG-Ia in an 11-month-old girl with developmental delay, failure to thrive, inverted nipples and abnormal fat pads.
View Article and Find Full Text PDF