Background Few scales are currently available to evaluate changes in hand volume. We aimed to develop a hand grading scale for quantitative assessments of dorsal hand volume with additional consideration of changes in skin texture; to validate and prove the precision and reproducibility of the new scale; and to demonstrate the presence of clinically significant differences between grades on the scale. Methods Five experienced plastic surgeons developed the Hand Volume Rating Scale (HVRS) and rated 91 images. Another five plastic surgeons validated the scale using 50 randomly selected images. Intra- and inter-rater agreement was calculated using the weighted kappa statistic and intraclass correlation coefficients (ICCs). Paired images were also evaluated to verify whether the scale reflected clinical differences. Results The intra-rater agreement was 0.95 (95% confidence interval, 0.922–0.974). The interrater ICCs were excellent (first rating, 0.94; second rating, 0.94). Image pairs that differed by 1, 2, and 3 grades were considered to contain clinically relevant differences in 80%, 100%, and 100% of cases, respectively, while 84% of image pairs of the same grade were found not to show clinically relevant differences. This confirmed that the scale of the HVRS corresponded to clinically relevant distinctions. Conclusions The scale was proven to be precise, reproducible, and reflective of clinical differences.

No abstract

No abstract

An 8-bit switched-capacitor multiply-and-accumulator (MAC) in 65nm CMOS is presented. Based on a cascaded low-power ring-amplifier-based switched-capacitor DACs, the MAC circuit features a programmable accumulation length in MAC computation. Fabricated in 65nm CMOS, the prototype MAC circuit achieves a precision-scaled energy efficiency of 1.32fJ per MAC operation, which is comparable to other state-of-the-art MAC circuits, along with best-in-class linearity. The noise performance has been verified using four real-world convolutional neural networks (CNNs) and 10,000-image data sets with up to 1,000 classes with an accuracy drop of less than 2% compared to the baseline 32-bit floating-point MAC.