An automated malaria cells detection from thin blood smear images using deep learning.

Timely and rapid diagnosis is crucial for faster and proper malaria treatment planning. Microscopic examination is the gold standard for malaria diagnosis, where hundreds of millions of blood films are examined annually. However, this method's effectiveness depends on the trained microscopist's skills. With the increasing interest in applying deep learning in malaria diagnosis, this study aims to determine the most suitable deep-learning object detection architecture and their applicability to detect and distinguish red blood cells as either malaria-infected or non-infected cells. The object detectors Yolov4, Faster R-CNN, and SSD 300 are trained with images infected by all five malaria parasites and from four stages of infection with 80/20 train and test data partition. The performance of object detectors is evaluated, and hyperparameters are optimized to select the best-performing model. The best-performing model was also assessed with an independent dataset to verify the models' ability to generalize in different domains. The results show that upon training, the Yolov4 model achieves a precision of 83%, recall of 95%, F1-score of 89%, and mean average precision of 93.87% at a threshold of 0.5. Conclusively, Yolov4 can act as an alternative in detecting the infected cells from whole thin blood smear images. Object detectors can complement a deep learning classification model in detecting infected cells since they eliminate the need to train on single-cell images and have been demonstrated to be more feasible for a different target domain.

An automated malaria cells detection from thin blood smear images using deep learning

@#Timely and rapid diagnosis is crucial for faster and proper malaria treatment planning. Microscopic examination is the gold standard for malaria diagnosis, where hundreds of millions of blood films are examined annually. However, this method’s effectiveness depends on the trained microscopist’s skills. With the increasing interest in applying deep learning in malaria diagnosis, this study aims to determine the most suitable deep-learning object detection architecture and their applicability to detect and distinguish red blood cells as either malaria-infected or non-infected cells. The object detectors Yolov4, Faster R-CNN, and SSD 300 are trained with images infected by all five malaria parasites and from four stages of infection with 80/20 train and test data partition. The performance of object detectors is evaluated, and hyperparameters are optimized to select the best-performing model. The best-performing model was also assessed with an independent dataset to verify the models’ ability to generalize in different domains. The results show that upon training, the Yolov4 model achieves a precision of 83%, recall of 95%, F1-score of 89%, and mean average precision of 93.87% at a threshold of 0.5. Conclusively, Yolov4 can act as an alternative in detecting the infected cells from whole thin blood smear images. Object detectors can complement a deep learning classification model in detecting infected cells since they eliminate the need to train on single-cell images and have been demonstrated to be more feasible for a different target domain.

[Evaluation of neoadjuvant chemo-radiotherapy with locally advanced rectal cancer by comparing tumour volume before and after treatment]. / Hodnocení neoadjuvantní chemoradioterapie u lokálne pokrocilého karcinomu rekta srovnáním objemu tumoru pred lécbou a po lécbe.

BACKGROUND: Neoadjuvant chemotherapy is a standard preoperative therapeutical procedure with locally advanced rectal adenocarcinoma. The aim of the study was to compare the tumour volume reduction before and after the oncological therapy in relation to the change in the CEA value and to the outcome of the histopathological evaluation of response to the treatment. PATIENTS AND METHODS: In the years 2004-2008, 274 rectal cancer patients were evaluated, of which 64 underwent neoadjuvant CRT with subsequent surgery and had also completed other inclusion criteria. The tumour volume before and after the CRT, percentage reduction in the tumour volume and the relation to the change in the CAE value and the histopathological evaluation were evaluated. RESULTS: The distance between the anus and the tumour was from 3 to 15 centimetres, the average value being 8.1 centimetres. In 5 cases the tumour was not histologically found in the resected specimen. Average value of the CEA value before the CRT was 18.12 ng/ml, range 0.7-98.1 ng/ml, after the CRT the average value was 7.00 ng/ml, range 0.5-18.7 ng/ml. The average tumour volume before CRT was 32.48, range 10.3-88.5, after the CRT the average volume was 20.13, range 4.7-55.1. CONCLUSION: A relation between the change in the T value and the volume reduction before and after the CRT of statistical significance has been proven in this group of patients. This relation however has not been proved in the N value change. Only in one-third of the evaluated patients was there a positive change in both T and N classification. No relation between the CEA value and the tumour volume change has been proven.

[Mirizzi syndrome with cholecystoduodenal fistula--a rare complication of cholecyctolithiasis]. / Mirizziho syndrom s cholecystoduodenální pístelí--vzácná komplikace cholecystolitiázy.

Mirizzi syndrome is a rare complication of the long-term gallbladder stone disease. It's caused by a gallbladder stone impression to the common bile duct wall. The clinical appearance is an obstructive jaundice, pain in right subchondrium and dyspepsia. Higher pressure in the extrahepatal bile ducts leads to biliobiliary or enterobiliary fistula formation. In the described case a cholecystoduodenal fistula was found. Preoperative diagnosis of MS is a medical challenge, and majority of the diagnosis is set introperatively. Surgery is the treatment of first choice. Ignoring the possibility of MS can seriously damage patient's health by injuring bile ducts during the surgery.

High-resolution profiles of trace metals in the pore waters of riverine sediment assessed by DET and DGT.

The techniques of DET (diffusive equilibrium in thin films) and DGT (diffusive gradients in thin films) were applied to obtain high-resolution vertical profiles of trace metals in freshwater sediments. In the framework of the EU-Interreg project Stardust (http://www.vliz.be/projects/stardust/) between France and Belgium, in which the mobility of sediment bound metals is investigated, sediment samples were collected from the Upper Scheldt River (at Helkijn, Belgium) and the Leie River (at Warneton, located at the Belgian-French border). Intra- and inter-laboratory comparisons of the gel techniques were carried out between the two laboratories involved. In general, a good agreement was observed, taking sediment heterogeneity into account. At both stations, metal pore water profiles show more or less similar tendencies although the sediment at Warneton was more anoxic than at Helkijn. A strong correlation between Fe and Co was found at Helkijn as well as at Warneton. The metal gradients at the water/sediment interface were calculated from the high resolution profiles and the conventional, low resolution profiles. Significant differences were observed.

Mercury depth profiles in river and marine sediments measured by the diffusive gradients in thin films technique with two different specific resins.

The diffusive gradients in thin films technique (DGT) was used to measure depth profiles of mercury in river and marine sediments in situ to a spatial resolution of 0.5 cm. Agarose gel was used as the diffusive gel in the DGT probes. Two different selective resins-Chelex 100 with iminodiacetic groups and Spheron-Thiol with thiol groups incorporated in the polyacrylamide resin gel-were tested. The different capture efficiencies of the two adsorbents enabled the fractions of mercury bound in different species in sediment pore water to be estimated. Mercury concentrations obtained by DGT with Spheron-Thiol resin were very similar to those obtained after centrifugation. This indicates that DGT with Spheron-Thiol resin reports on total dissolved mercury levels. The concentration of mercury measured by DGT with Chelex-100 resin was much lower (by a factor of 5-20) for the same sediment samples. Chelex-100 does not have such a high affinity to mercury as Spheron-Thiol, and so it only reports on the content of labile mercury species, such as inorganic ions and weak complexes. The content of labile mercury species in the river sediment was approximately 20% of the total dissolved mercury in pore water, whereas in marine sediment only 7% of the mercury was present as labile species.

Application of diffusive gradient in thin films technique (DGT) to measurement of mercury in aquatic systems.

The diffusive gradient in thin films (DGT) technique was investigated and used to measure mercury concentration in river water. Mercury ions are covalently bound to amide nitrogen groups of commonly used polyacrylamide, which makes this gel unsuitable as a diffusive medium. In contrast, agarose gel was found as the diffusive gel for mercury measurements. Basic performance tests of agarose DGT verified the applicability of Fick's first law for DGT measurements. Two selective resins, Chelex-100 with iminodiacetic groups and Spheron-Thiol with thiol groups were used. The measured diffusion coefficient in agarose gel was close to that in water. The concentration of mercury in Svitava river measured by DGT with Speron-Thiol resin gel was higher (0.0116 +/- 0.0009mugl(-1)) than those obtained by Chelex-100 (0.0042 +/- 0.0005mugl(-1)). Different capture efficiencies of two adsorbents enable to estimate fractions of mercury bonded in different complexes in the river water. The concentrations of mercury found by DGT both Chelex-100 and Speron-Thiol resin gels are much lower than that measured directly in the river water (0.088 +/- 0.012mugl(-1)). This difference indicates that DGT concerns inorganic ions and labile species only, and that it is not able to include inert organic species and colloids.

[Dynamics of the sliding screw]. / Dynamický skluzný sroub.

Fractures of the upper end of the femur are treated by the authors since August 1993 by the method of a dynamic gliding screws. This method is considered sparing and economical. The operation is implemented as urgent with a screen of antibiotics and anticoagulants in order to reduce postoperative complications. As a matter of routine intensive early rehabilitation is ensured and early return of the patient to his normal environment. Between August 1, 1993 and August 31, 1998 at the authors clinic 259 osteosyntheses were made using a dynamic gliding screw. There were 10 reoperations. In three instances the reason was incorrect surgical technique, in seven instances unsuitable indication of osteosynthesis in an arthrotic joint with impaired nurture in a severely dislocated fracture of the subcapital type. Reoperation was implemented twice by the original technique, eight times by replacement of the hip joint [4, 6]. For successful treatment of patients with fractures of the upper end of the femur it is essential that the department should have an X-ray amplifier, instruments for a dynamic gliding screw, a gamma and reconstruction nail and a hip joint replacement.