A new computerized method of reading sperm morphology (strict criteria) is as efficient as technician reading

Abstract

Objectives: To compare the ability of a computerized method of sperm morphology with the manually recorded method in predicting in vitro fertilization (IVF) results, to compare results obtained by both methods, and to determine the intraobservation variability.

Design, setting, patients: Forty-three stained semen slide preparations from two large level-three academic institutions' reproductive endocrinology units (IVF programs) were blindly evaluated, and the sperm were classified into normal and amorphous forms.

Results: Experiment 1: Twenty-one slide preparations from the Tygerberg gamete intrafallopian transfer program were manually evaluated; the fertilization rates for the groups with < 14% and > 14% normal sperm forms were 33.3% (15/45 oocytes) and 76.6% (46/60 oocytes), respectively. Corresponding fertilization rates with FERTECH were 46.8% (30/64) and 75.6% (31/41). Experiment 2: Twenty-two slide preparations from the Norfolk IVF program were evaluated. The manual method reported a fertilization rate in the group with < 14% normal forms of 27.4% (14/51 oocytes) compared with 90.0% (127/141 oocytes) in the group with > 14% normal forms. Corresponding figures for the FERTECH method were 33.9% (18/53) and 88.4% (123/139), respectively. Experiment 3: When the 43 slide preparations were blindly evaluated using both methods, 84% of the FERTECH evaluations correlated well with the manual method and FERTECH ability to diagnose the subfertile male (< 14% normal forms) was 95% (sensitivity). Experiment 4: A total of 16 different slides (8 per group) were randomly selected and analyzed five times (100 cells per reading) by the computerized method. The slides were obtained from men with normal sperm morphology of < 14% and > 14% as classified by the manual method. In the first group (< 14%) 97.5% (39/40) of the readings classified the sperm in the proper category, whereas in the second group (> 14%) 95% (38/40) of the cases were correctly identified.

Conclusion: Using strict criteria for morphology evaluation, there is a positive and significant correlation between FERTECH evaluation and manual assessment. The reproducibility of the computerized method and the ability to distinguish between fertile and subfertile groups using those criteria are good.