Veterinary clinical pathology最新文献

Background: The empirical biologic variation (EBV) method to derive analytical performance specifications (APSs) has been provisionally evaluated for use in the veterinary clinical pathology laboratory. Analytical performance can affect EBV APSs. It has been argued that this effect is minimal, but the effect has not been evaluated for the range of performance found in veterinary laboratories.

Objective: Model and compare EBV APSs derived across a range of analytical CV (CV_A) observed in veterinary clinical pathology laboratories.

Method: Published data on biochemistry measurand CV_A in veterinary clinical pathology laboratories and dog reference intervals (RIs) were used to model EBV APSs using Monte Carlo simulation. The modeled EBV APSs were compared with analogous APSs from total allowable error (TE_A) and traditional BV models.

Results: Generally, the EBV-recommended permissible CV_A (pCV_A) was smaller than the observed CV_A when analytical performance was poor. The EBV APSs did not have a consistent relationship with analogous APSs from TE_A or traditional BV systems when analytical performance was judged acceptable using TE_A or BV APSs. When analytical performance was unacceptable, the resulting EBV APSs increased dramatically (sometimes by > 200%) and lost their initial relationship with TE_A or traditional BV-derived APSs. Using log-normal calculations for measurands with a Gaussian distribution produced wider EBV APSs.

Conclusion: The EBV method produces growth-oriented APSs under all modeled conditions and can produce notably more lenient APSs for poorly performing labs than better performing labs. The EBV method can produce acceptable APSs under some conditions and may not be suitable for use in veterinary medicine.

Artificial intelligence (AI) is emerging as a valuable diagnostic tool in veterinary medicine, offering affordable and accessible tests that can match or even exceed the performance of medical professionals in similar tasks. Despite the promising outcomes of using AI systems (AIS) as highly accurate diagnostic tools, the field of quality assurance in AIS is still in its early stages. Our Part I manuscript focused on the development and technical validation of an AIS. In Part II, we explore the next step in development: external validation (i.e., in silico testing). This phase is a critical quality assurance component for any AIS intended for medical use, ensuring that high-quality diagnostics remain the standard in veterinary medicine. The quality assurance process for evaluating an AIS involves rigorous: (1) investigation of sources of bias, (2) application of calibration methods and prediction of uncertainty, (3) implementation of safety monitoring systems, and (4) assessment of repeatability and robustness. Testing with unseen data is an essential part of in silico testing, as it ensures the accuracy and precision of the AIS output.

Background: Manual blood smear review (MSR) to complement automated CBC results is a labor-intensive process. Efforts have been made to use criteria based on automated hematology analyzer data to identify samples warranting MSR. These efforts have coincided with the emergence of modern data science and machine learning.

Objective: In this study, we aim to determine if machine learning can reduce manual smear review (MSR) rates while meeting or exceeding the performance of traditional MSR criteria.

Method: 9938 automated CBCs with paired MSRs were performed on samples from rhesus and cynomolgus macaques. The definition of a positive (abnormal) smear was determined. Two expert-derived MSR criteria were created: criteria adapted from published, standardized human laboratory criteria (Adapted International Consensus Guidelines[aICG]) and internally generated criteria (Center Consensus Guidelines [CCG]). An ensemble machine learning model was trained on an independent subset of the data to optimize the balanced accuracy of classification, a combined measure of sensitivity and specificity. The resulting machine learning model and the two expert-derived MSR criteria were applied to a test dataset, and their performance compared.

Results: aICG criteria demonstrated high sensitivity (80.8%) and MSR rate (74.2%) while CCG criteria demonstrated lower sensitivity (57.1%) and MSR rate (36.1%). The machine learning model integrated with CCG criteria had a superior combination of both sensitivity (76.8%) and MSR rate (45.1%) achieving a false negative rate of 1.6%.

Conclusion: Machine learning in combination with expert-derived criteria can optimize the selection of samples for MSR thus decreasing MSR rates and labor efforts required for CBC performance.

Background: Lymph node (LN) aspirates are often obtained to distinguish large-cell lymphoma (LCL) from non-lymphoma (NL) in dogs with enlarged lymph nodes.

Objective: Images from cytology slides tested the effects of magnification, image type, and number on a convolutional neural network (CNN) differentiating canine LCL from NL.

Methods: Three hundred images of LCL and NL were used to train a CNN and interrogate the effects of image magnification, type, and number on the model's performance. Identified cases were imaged at 200×, 500×, and 1000× magnification in color and gray-scale and then used to train and identify optimal magnification and image type. The impact of the image number per cohort (50, 100, 150, 200, 250, 300) on the top model's performance was then assessed.

Results: The highest performance with color images was achieved at 1000× magnification, with an accuracy of 95.8%, a Receiving Operating Characteristic (ROC) area of 0.997, and an F-measure of 0.958. Similarly, the best results with gray images, also at 1000× magnification, showed an accuracy of 96.67%, a ROC area of 0.994, and an F-measure of 0.967. Performance improvements were most significant and plateaued as the number of images per class approached 150, with an accuracy of 95%, ROC area of 0.939, and F-measure of 0.95.

Conclusion: The analysis across models suggests that color versus greyscale did not significantly impact overall performance to distinguish LCL or NL. Optimal magnification was 1000×. A minimum of 150 images per class is recommended for pilot CNN studies in this 2-class problem.

A 3-year-old, intact female Havanese dog was presented for treatment of a relapse of suspected immune-mediated anemia and thrombocytopenia. Over the course of treatment, the patient's anemia appeared to acutely transition from macrocytic and hypochromic to microcytic and hypochromic between days five and six of hospitalization. Assessment of red blood cell volume/hemoglobin content (RBC V/HC) cytograms produced on the ADVIA 2120i (Siemens Healthcare Diagnostics, Munich, Germany) revealed a new population of microcytic and hypochromic erythrocytes on day six of hospitalization. This finding prompted the consideration that the microcytic and hypochromic erythrocyte population was introduced to the patient via a packed red blood cell (pRBC) transfusion, which was confirmed after running a sample of pRBCs from the same donor through the ADVIA 2120i. Other causes of microcytosis, such as iron deficiency anemia, portosystemic shunt, and hyponatremia, were additionally investigated and considered less likely. The cause of the microcytosis in the donor was ultimately undetermined. To the authors' knowledge, this is the first case where inadvertent transfusion with microcytic pRBCs was discovered with the aid of RBC V/HC cytograms. This case highlights the importance of considering pRBCs as a source of changes in erythrocyte indices as well as the utility of RBC V/HC cytograms in patients with multiple populations of erythrocytes, such as those receiving transfusions. Additionally, this case prompts reconsideration of the need for standardization of guidelines for blood donation frequency and monitoring of iron status in frequent blood donors.

Background: Measurement of liver enzymes and metabolites for both clinically healthy and sick cattle is a routine part of dairy herd management. Reference intervals (RIs) are influenced by many variables, including pregnancy, breeding, and geographical variables, and can shift over time. Few previous studies have addressed the specific RIs of dairy cows, and none have specifically addressed the RIs of German Holstein Friesian cows.

Objective: The aim was to determine the RIs of German Holstein Friesian cows considering variables such as age, parity, milk yield, body condition score, days in milk, and pregnancy. Additionally, this study aimed to evaluate the effects of anticoagulant use on the RIs.

Methods: Serum, lithium heparin, and EDTA plasma samples from 131 lactating, apparently healthy Holstein Friesian cows from 10 dairy farms were collected. The levels of glutamate dehydrogenase (GLDH), γ-glutamyltransferase (GGT), aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), creatine kinase (CK), cholesterol, nonesterified fatty acids (NEFAs), and total bilirubin were measured, and new RIs were determined.

Results: The following RIs were determined: beta-hydroxybutyrate (BHB): 0.25-1.00 mmol/L; ALP: 71.1-258.9 U/L; GLDH: 9.1-121.0 U/L; CK: 65.4-257.2 U/L; GGT_pregnant: 8.7-65.3 U/L; GGT_{not pregnant}: 12.3-48.6 U/L; NEFAs_{> 42 Day in Milk}: 65.4-308.7 μmol/L; ALT: 14.0-34.5 U/L; cholesterol: 2.7-7.5 mmol/L; AST_pregnant: 76.6-279.7 U/L; AST_{not pregnant}: 67.1-187.1 U/L; total bilirubin: 0.9-5.2 μmol/L.

Conclusion: New and more precise RIs for cattle could help veterinarians detect hepatocyte damage with minor enzyme leakage and liver stress at an early stage.

Artificial intelligence (AI) has transformative potential in veterinary pathology in tasks ranging from cell enumeration and cancer detection to prognosis forecasting, virtual staining techniques, and individually tailored treatment plans. Preclinical testing and validation of AI systems (AIS) are critical to ensure diagnostic safety, efficacy, and dependability. In this two-part series, challenges such as the AI chasm (ie, the discrepancy between the AIS model performance in research settings and real-world applications) and ethical considerations (data privacy, algorithmic bias) are reviewed and underscore the importance of tailored quality assurance measures that address the nuances of AI in veterinary pathology. This review advocates for a multidisciplinary approach to AI development and implementation, focusing on image-based tasks, highlighting the necessity for collaboration across veterinarians, computer scientists, and ethicists to successfully navigate the complex landscape of using AI in veterinary medicine. It calls for a concerted effort to bridge the AI chasm by addressing technical, ethical, and regulatory challenges, facilitating AI integration into veterinary pathology. The future of veterinary pathology must balance harnessing AI's potential while intentionally mitigating its risks, ensuring the welfare of animals and the integrity of the veterinary profession are safeguarded. Part I of this review focuses on considerations for model development, and Part II focuses on external validation of AI.

Background: Hemorheology is used in human medicine but remains largely unexplored in veterinary medicine. The assessment of RBC rheology holds potential for diagnosing, monitoring, and predicting outcomes in veterinary species. Reference intervals (RIs) are needed before clinical assessment of the method.

Objectives: Determine RIs for hemorheological parameters in dogs, compare RIs generated at 620 versus 800 nm wavelengths, as well as RIs generated with and without following ASVCP guidelines.

Methods: Whole blood was sampled on EDTA tubes from healthy dogs and run on the MIZAR Analyzer for 13 parameters (DI, Int (d), PEAK, AI, Int (a), AMP, BpL, Bp, BpH, T½, OTF, OT, ED). After outlier removal considerations, the normality of distribution was assessed, and RIs were calculated using parametric or non-parametric methods (ASVCP guidelines). RIs performed at 620 and 800 nm, with and without following ASVCP guidelines, were compared.

Results: RIs were established for rheology parameters at 620 and 800 nm in a population of 42 healthy dogs. 5/13 parameters had nearly identical RI limits between 620 and 800 nm; others showed an overt shift at 800 nm toward higher values for both RI limits. Reference intervals showed similar results whether ASVCP guidelines were followed or not for 7/13 parameters, namely when data had a Gaussian distribution and contained a maximum of one outlier. For the remaining parameters, RIs calculated without following ASVCP guidelines differed significantly and were considered unsuitable for clinical use.

Conclusions: Specific RIs were established for 620 and 800 nm wavelengths. Compliance with ASVCP guidelines was deemed necessary.

Background: Population-based reference intervals (RIs) are fundamental for appropriate result interpretation. Current recommendations for their determination specify that a direct, de novo approach is preferable, including at least 120 carefully selected subjects. This is not always feasible due to logistical and financial constraints. Recently, several sophisticated algorithms for the indirect estimation of RIs have been proposed in human medicine and described as comparable with direct methods in many cases.

Objectives: Preliminary evaluation of the utility and feasibility of novel indirect methods for determining RIs in the veterinary clinical laboratory.

Methods: Calcium and alanine aminotransferase (ALT) were selected as representative measurands expected to show Gaussian and non-Gaussian distributions, respectively. After limited data filtering, canine results extracted from a laboratory database were used to estimate indirect RIs and 90% confidence intervals (CIs) using the RefineR algorithm. Using 90% CIs as well as permissible differences (pD), the indirect RIs were compared to direct RIs derived by the same laboratory, over the same period, from a population of healthy canine blood donors.

Results: The direct and indirect approaches yielded comparable reference limits for both measurands. Apart from the upper reference limit for Ca, 90% CIs for the reference limits calculated from the conventional direct and indirect methods overlapped. Direct reference limits were within pD calculated from the indirect reference limits.

Conclusion: The indirect method, using the RefineR algorithm, may represent a cost-effective option for RI determination. It may also complement direct approaches and has the potential to be more representative of animal populations.