To evaluate the accuracy of predicting refractive outcomes of swept-source optical coherence tomography based biometry (ARGOS; Movu Inc., Santa Clara, CA, USA) in nuclear cataracts. Methods

A total of 107 eyes (107 nuclear cataract patients) were analyzed. Subjects were divided into three groups according to the maximum nuclear density of Pentacam HR (A, lower tertile; B, medium tertile; C, upper tertile). The keratometry and axial length measured by IOLMaster (Carl Zeiss Meditec, Jena, Germany) and ARGOS systems were compared for each group. The correlation between maximum nuclear density and axial length difference readings from the two instruments was evaluated. The mean absolute error between the predicted refraction and 2-month post-operative refraction was compared. Results

The maximum nuclear densities were 28.31 ± 7.30, 51.37 ± 7.82, and 88.63 ± 11.23 for groups A, B, and C, respectively. The axial length measured by ARGOS was significantly longer than that obtained using IOLMaster for groups B and C (respectively, p = 0.035, p < 0.001). A significantly positive correlation was found between the maximum nuclear density and axial length difference of the two devices ( p < 0.001). Mean absolute errors were not significantly different between IOLMaster and ARGOS in group A. However, in groups B and C, the mean absolute error using ARGOS (0.31 ± 0.22 D and 0.32 ± 0.20 D, respectively) was significantly lower than that of IOLMaster (0.43 ± 0.21 D and 0.50 ± 0.26 D, respectively) (Group B, p = 0.027; Group C, p = 0.001). Conclusions

Even in dense nuclear cataracts, accurate refractive outcome prediction was possible using swept-source optical coherence tomography based biometry.