RESUMEN
Deep learning (DL) models for medical image classification frequently struggle to generalize to data from outside institutions. Additional clinical data are also rarely collected to comprehensively assess and understand model performance amongst subgroups. Following the development of a single-center model to identify the lung sliding artifact on lung ultrasound (LUS), we pursued a validation strategy using external LUS data. As annotated LUS data are relatively scarce-compared to other medical imaging data-we adopted a novel technique to optimize the use of limited external data to improve model generalizability. Externally acquired LUS data from three tertiary care centers, totaling 641 clips from 238 patients, were used to assess the baseline generalizability of our lung sliding model. We then employed our novel Threshold-Aware Accumulative Fine-Tuning (TAAFT) method to fine-tune the baseline model and determine the minimum amount of data required to achieve predefined performance goals. A subgroup analysis was also performed and Grad-CAM++ explanations were examined. The final model was fine-tuned on one-third of the external dataset to achieve 0.917 sensitivity, 0.817 specificity, and 0.920 area under the receiver operator characteristic curve (AUC) on the external validation dataset, exceeding our predefined performance goals. Subgroup analyses identified LUS characteristics that most greatly challenged the model's performance. Grad-CAM++ saliency maps highlighted clinically relevant regions on M-mode images. We report a multicenter study that exploits limited available external data to improve the generalizability and performance of our lung sliding model while identifying poorly performing subgroups to inform future iterative improvements. This approach may contribute to efficiencies for DL researchers working with smaller quantities of external validation data.
RESUMEN
Plastic additives are widely used in plastic production and are found in the environment owing to their widespread applications. Among these additives, N-butyl benzenesulfonamide (NBBS) and triphenyl phosphate (TPHP) are under international watchlist for evaluation, with limited studies on amphipods. Di-ethylhexyl phthalate (DEHP) and dibutyl phthalate (DBP) are banned in some countries and categorised as substances of very high concern. This study aimed to investigate the effects of NBBS, TPHP, DEHP and DBP on the swimming activity of a coastal intertidal marine amphipod, Echinogammarus marinus. Furthermore, this study is the first to quantify startle response in E. marinus in response to light stimuli. Amphipods were exposed to 0, 0.5, 5, 50 and 500 µg/l concentrations of all test compounds. Swimming activity and startle responses were assessed by video tracking and analysis using an 8-min alternating dark and light protocol after exposure on days 7 and 14. We observed an overall compound and light effect on the swimming activity of E. marinus. A significant decrease in swimming distance was found in 500 µg/l NBBS and TPHP. We observed that the startle response in E. marinus had a latency period of >2 s and animals were assessed at 1 s and the sum of the first 5 s. There was a clear startle response in E. marinus during dark to light transition, evident with increased swimming distance. NBBS exposure significantly increased startle response at environmental concentrations, while significant effects were only seen in 500 µg/l TPHP at 5 s. We found no significant effects of DEHP and DBP on swimming behaviour at the concentrations assessed. The findings of this study affirm the necessity for a continuous review of plastic additives to combat adverse behavioural effects that may be transferable to the population levels.