A combined deep learning and radiomics model helps clinicians distinguish between the complex and the simple Acute appendicitis (AA) Researchers discovered this using pelvic CT.
The study results may help clinicians better decide whether appendicitis patients can be managed non-surgically, said Dan Liang of the First Affiliated Hospital of Jinan University in Guangzhou, Guangdong Province, China. A team led by Dr. The research results were announced on September 27th. academic radiology.
The group notes that “appendectomy is one of the most frequently performed surgeries worldwide.” “However, it is associated with short- and long-term complications and represents a significant burden on modern global public health systems… [Now] Non-surgical management of AA is [begun to be considered]. ”
The group noted that acute appendicitis is one of the most common causes of acute abdominal pain. However, debate continues as to whether surgery is actually necessary in each case of appendicitis.
Over the past 30 years, several randomized controlled trials have shown that complex AA (gangrenous or perforating) and simple AA are two distinct entities with different pathophysiology, and that the latter as an initial treatment strategy. The researchers explained that the use of antibiotics “should be a safe option. Accurate diagnosis and evaluation and selection of the most appropriate treatment options in patients with AA are therefore of paramount importance.” .
Liang et al. conducted a study of 11,765 adult patients with acute appendicitis. Of these, 700 were included in the training cohort for the deep learning algorithm (which the team developed using his CatBoost) and 465 were included in the validation cohort. This model incorporates clinical characteristics and her CT visual features, deep learning features, and radiomics features. In this study, we used the area under the receiver operating characteristic curve (AUROC) measurement (with 1 as the criterion) to compare the performance of the model to a radiologist’s interpretation of her CT examination. Surgical and pathology records were used as the reference standard to determine whether acute appendicitis was complicated or uncomplicated. The group also assessed accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
A model combining deep learning and radiomics showed higher AUROC and higher sensitivity and NPV than radiologist readers.
| Validation set data comparing a model combining deep learning and radiomics with radiologist readings to differentiate between complex and simple acute appendicitis | ||
|---|---|---|
| measurement | Radiologist’s diagnosis | A model that combines deep learning and radiomics |
| AUC (based on 1) | 0.68 | 0.79 |
| Accuracy | 73% | 72% |
| sensitivity | 45% | 70% |
| specificity | 90% | 74% |
| PPV | 75% | 63% |
| NPV | 73% | 80% |
”[Our] The composite model allows to accurately distinguish between complex and simple AA, providing a significant advantage compared to the radiologist’s visual diagnosis.
“The aim is to improve the process of patient selection for non-surgical management,” the group concluded.
You can find the full research results here.