Bone age determination using only the index finger: a novel approach using a convolutional neural network compared with human radiologists

Background

Recently developed convolutional neural network (CNN) models determine bone age more accurately than radiologists.

Objective

The purpose of this study was to determine whether a CNN and radiologists can accurately predict bone age from radiographs using only the index finger rather than the whole hand.

Materials and methods

We used a public anonymized dataset provided by the Radiological Society of North America (RSNA) pediatric bone age challenge. The dataset contains 12,611 hand radiographs for training and 200 radiographs for testing. The index finger was cropped from these images to create a second dataset. Separate CNN models were trained using the whole-hand radiographs and the cropped second-digit dataset using the consensus ground truth provided by the RSNA bone age challenge. Bone age determination using both models was compared with ground truth as provided by the RSNA dataset. Separately, three pediatric radiologists determined bone age from the whole-hand and index-finger radiographs, and the consensus was compared to the ground truth and CNN-model-determined bone ages.

Results

The mean absolute difference between the ground truth and CNN bone age for whole-hand and index-finger was similar (4.7 months vs. 5.1 months, P=0.14), and both values were significantly smaller than that for radiologist bone age determination from the single-finger radiographs (8.0 months, P<0.0001).

Conclusion

CNN-model-determined bone ages from index-finger radiographs are similar to whole-hand bone age interpreted by radiologists in the dataset, as well as a model trained on the whole-hand radiograph. In addition, the index-finger model performed better than the ground truth compared to subspecialty trained pediatric radiologists also using only the index finger to determine bone age. The radiologist interpreting bone age can use the second digit as a reliable starting point in their search pattern.

     

Successful biventricular cardiac resynchronization therapy in a failing Fontan patient: Implications of ventriculo‐ventricular interdependency in single ventricle physiology

We describe a Fontan patient with severe heart failure who was successfully treated with biventricular cardiac resynchronization therapy (CRT). Our case shows that strain imaging might play a crucial role in guiding placement of pacing leads and in characterizing the electromechanical substrate associated with a favorable CRT response. Furthermore, we demonstrate for the first time that ventriculo‐ventricular interdependency seems an important mechanical concept, which can be utilized to augment cardiac performance in failing Fontan patients with a functional hypoplastic ventricle.     

Treatment Fidelity of a Nurse-Led Motivational Interviewing-Based Pre-Treatment in Pain Rehabilitation

Treatment fidelity and proficiency of a nurse-led motivational interviewing (MI)-based pre-treatment and control condition was evaluated. A random sample was scored by means of the Motivational Interviewing Treatment Integrity (MITI) scale, and a second rater was in charge. MI fidelity was satisfactory for three out of five ratings. Most mean ratings were higher in the MI-based intervention, but differences were not statistically significant. The threshold for beginning MI proficiency was only exceeded for one score and one additional measure. In general, higher levels of fidelity in the intervention condition confirmed that MI was partially applied there. Although the quality of MI delivery as well as mixed inter-rater reliabilities of the fidelity scores leaves room for improvement, robust findings between the two raters were found. These results suggest the need for rigor selection of MI counselors on beforehand, and continuous supervision. Furthermore, fidelity check in studies using MI is needed.