# Novel height estimation formula that accounts for the effects of aging based on lumbar length measurements in postmortem CT images

## TL;DR

This paper introduces a new height estimation formula using lumbar spine measurements from CT scans that is not affected by aging, improving accuracy for elderly individuals.

## Contribution

A novel height estimation formula based on lumbar spine length that accounts for age-related shortening in the elderly.

## Key findings

- Lumbar spine indices like SPLV and LVTL showed strong correlations with height (R = 0.762 and 0.761).
- The new formula using PLV2, SPLV, and LVTL is not influenced by age, unlike traditional sternal length-based methods.
- The formula achieved standard errors of estimation as low as 4.448 cm for LVTL-based estimates.

## Abstract

In forensic practice, personal identification using expert testimony is important in unidentified cases, and the height of the deceased is indispensable in identification. Although many height estimation formulas have been reported, height estimates are often too great in the elderly due to age-related shortening. In this study, we address this problem by developing a height estimation formula based on measurement of the lumbar spine, which is thought to shorten with age. To develop a height estimation equation based on lumbar spine length, 183 postmortem CT images taken at our institute from 2016 to 2023 (ages 19–95 years) were prepared for the training dataset and 78 images were used for the validation dataset. In all training dataset cases, anterior margin height (ALV), central height (CLV), and posterior margin height (PLV) of each lumber vertebra and total lumber spine length including the intervertebral disc (LVTL) were measured on 3D CT-reconstructed images. The sum of the ALV (SALV), CLV (SCLV), and PLV (SPLV) of all lumbar vertebrae were calculated by image analysis software, and the correlation between each index and height was examined. As a control, an estimation equation based on sternal length was developed. Significant positive correlations were observed between each of the lumbar spine indices and height, with the PLV of the second lumbar vertebrate (PLV2) (R = 0.710), SPLV (R = 0.762), and LVTL (R = 0.761) showing the strongest correlation. R2 and standard error of estimation (SEE) were 0.622 and 4.926 cm for PLV2, 0.683 and 4.515 cm for SPLV, and 0.692 and 4.448 cm for LVTL, respectively. Furthermore, estimation equations based on sternal length often estimated higher values for elderly persons and did not take into account the effect of aging, while those based on PVL2, SPLV, and LVTL showed no correlation with age. In conclusion, we consider that our new formula for estimating height based on lumbar spine length, especially on PLV2, SPLV, and LVTL, is not affected by aging.