Hookworm infection remains of considerable importance to public health. However, because critical cases caused by hookworm infection are rarely observed in China, accurate and prompt diagnoses are difficult to achieve in clinical practice. In this study, we describe how we combined morphological and molecular approaches to achieve the clinical diagnosis of hookworm infection.

A 75-year-old Chinese woman who presented with dizziness, poor appetite, poor sleep, and weakness in her limbs was diagnosed with chronic atrophic gastritis and was positive for Helicobacter pylori, iron deficiency anemia with a hemoglobin concentration of 35 g/L, and left atrial enlargement. However, after symptomatic treatment, the patient did not improve. Upper gastrointestinal endoscopy revealed the presence of live nematodes in the descending portion of the patient's duodenum. Fecal examination via saturated brine flotation revealed hookworm eggs. Further verification via semi-nested reverse transcription-polymerase chain reaction assay confirmed provided confirmation that the hookworm species was Necator americanus. Albendazole was used for antihelminthic treatment. Through follow-up visits, we found that the antihelminthic treatment was successful and that her anemia was cured.

In this study, a combination of morphological and molecular approaches were used to make a definite diagnosis of severe iron deficiency anemia caused by Necator americanus infection in a patient. The results presented here provide suitable guidance for the clinical diagnosis of hookworm infection and a powerful tool for the identification of hookworms.

To evaluate the degree of gastric, enteric, colonic, and rectal filling in multidetector computed tomography (MDCT) whole gastroenterography.

In this prospective study involving 124 patients, 78 and 46 patients underwent MDCT whole gastroenterography using positive and neutral oral contrast agents, respectively. The degree of filling of the stomach, small and large bowel, was qualitatively analyzed by experienced radiologists using a 3-point scoring system.

The majority of patients received a score of ≥2 for small intestine filling using both positive and neutral contrast agents (90.5% and 78.2%, respectively), and <9% of the patients had a score of 0. The highest score for the degree of filling in the small intestine was observed in the ileum, followed by the duodenum and jejunum. There was a significant difference in the degree of filling achieved with positive and neutral contrast agents in the duodenum (P = .013) and jejunum (P = .047). More than 74% of cases had an optimal filling of the stomach, whereas >80% of the cases had an optimal filling of the colorectal segments. Only ≤5.1% had a score of 0 for the analyzed segments of the colorectum. Positive and neutral contrast agents were associated with similar degree of filling in the stomach and colon segments without a significant difference in the extent of contrast agent filling (P > .05).

Multidetector computed tomography whole gastroenterography was found to be a simple, safe, noninvasive, painless, and effective modality for the diagnosis of stomach and bowel complications in clinical settings.

The success of Convolutional Neural Network (CNN) is attributed to their ability to learn rich midlevel image representations as opposed to hand-crafted low-level features used in many natural image classification methods. Learning CNN, however, amounts to estimating millions of parameters and requires a very large number of annotated image samples. In this paper, we explored transfer learning for gastrointestinal bleeding detection on small-size imbalanced endoscopy images, and showed how image representations learned with CNN on large-scale annotated datasets can be efficiently transferred to other tasks with limited amount of training data. We first transferred pre-trained Inception V3 model trained on the ImageNet dataset to compute mid-level image representation, and then fine-tuned the trained model with labeled endoscopy images, and resumed training from already learned weights. Additionally, we introduce both data augmentation and image resampling to increase the size of the training database and the positive sample rate to perform the Transfer Learning. Our results showed that our transfer learning method produces the best performance on AUC (the area under the receiver operating curve), Precision, Recall and Accuracy as compared to both the hand-crafted feature based method and training CNN model from-scratch method.