The potential of measuring brain tissue mechanics for forensic time since death estimations was recently demonstrated [1, 2]. However, the approach showed limited resolution in time since death prediction after 24 h. Screening for notable differences between day 0 (‘fresh samples’) and the subsequent testing days is of paramount importance. Comparisons between different testing days beyond day 0, such as between days 1 and 2, are valuable for detecting points in time with marked biomechanical changes. However, they offer limited practical value for forensic time since death estimations. When storing the brains at 20 °C between death and the mechanical testing, the most prominent differences were observed between fresh samples and samples with a PMI of one day [2]. As a result, day 0 samples could be distinguished from samples with a PMI of at least one day with high sensitivity and specificity [2]. This study used tissue preparation and testing conditions identical to our previous study [2]. The only change was the storage temperature, which was lowered to 4 °C immediately after retrieval of the brains following animal sacrifice.

At a post-mortem storage temperature of 20 °C, the FL, ADB, CB, and SC significantly differed from the day 0 values after one day post-mortem in all investigated biomechanical properties [2]. Three days post-mortem, all brain regions examined in the study significantly differed from the day 0 values at 20 °C in all biomechanical properties [2]. At 4 °C, the biomechanical changes were much slower. Sample storage at 4 °C for one day did not significantly alter the investigated biomechanical properties of brain tissue. This implies that biomechanical analyses for time since death estimations of brain tissue could be delayed for at least 24 h, potentially even 48 h, by maintaining constant cooling at 4 °C. In practice, this could enable forensic investigators to delay biomechanical analyses for at least a couple of hours, for instance, when other tasks must be prioritized. Beyond that, the tissues could be transferred to a specialized laboratory in a moist 4 °C environment if the forensic team on site is unable to perform the biomechanical analyses immediately. However, it remains unclear if the delayed onset of degradation observed only applies when the tissues are transferred to a 4 °C environment in a fresh state or if it also occurs if initiated at a later stage. This question remains to be addressed in future research.

The course of the curves for the investigated biomechanical properties was examined to determine whether general trends apply among the different sampling regions. However, the curves’ progression was region-specific, likely linked to underlying structural differences, including variations in cell types and the white-to-grey matter ratio.

As observed at 20 °C, the CB samples exhibited a distinct drop in the investigated biomechanical properties, followed by a slow decline to reach a plateau [2]. At 4 °C, this significant drop was delayed by one day compared to storage at 20 °C. However, it is important to note that cool storage does not imply a simple one-day time delay for biomechanical changes observed at 20 °C storage. For instance, the FL and ADB samples were among the earliest to differ from day 0 samples after just one day post-mortem at 20 °C [2], but at 4 °C, they were among the more stable sample regions.

From a practical perspective, forensic investigators require cut-off values to estimate the time since death with high diagnostic ability. Broadly, the diagnostic ability of ROC curves based on the AUC can be categorized as random (AUC: 0.5–0.6), poor (0.6–0.7), fair (0.7–0.8), good (0.8–0.9), and excellent (0.9–1.0) [25]. However, ROC analyses should be interpreted beyond AUC labels [25]. This study revealed that, when using CB samples stored at 4 °C, a PMI of at least two days could be determined with excellent diagnostic ability when reaching CSmod values below 1435 Pa or Smod values below 1313 Pa. A PMI of at least three days could be determined with good diagnostic ability using a Lmod cut-off value of 415 Pa for the CB. For a PMI of at least four days, a good diagnostic ability was achieved using a CSmod cut-off value of 1656 Pa for the SC. In conclusion, the hypothesis that cooling at 4 °C extends the value of biomechanical time since death analyses beyond day one after death, compared to storage at 20 °C, can be accepted. However, in contrast to the results at 20 °C, at 4 °C, it was impossible to differentiate samples with a PMI of one day from fresh samples. Therefore, cooling not only extends the value of biomechanical time since death estimations but also lowers the diagnostic value of the method in the very early post-mortem phase, as the biomechanical properties are kept stable. Importantly, the diagnostic abilities stated above require a constant sample temperature for the given intervals. From a practical standpoint, the tissue temperature post-mortem usually varies over time due to different external influences such as exposure to sunlight or the transfer of body parts between places by the perpetrator. Future investigations should explore to what extent the given diagnostic abilities stated here are lowered under varying degrees of temperature variation throughout the given time frames.

Although the diagnostic ability of the CSmod of the SC samples for PMIs of at least four days was good, further analyses are warranted due to the limited number of samples that fell below the cut-off value of 1656 Pa in the given dataset. None of the investigated biomechanical properties of the deep brain samples showed significant differences from day zero within the four testing days post-mortem at a storage temperature of 4 °C. Therefore, future studies should consider extending the analyzed PMI interval of the ADB and PDB at 4 °C to determine their points in time of marked tissue degradation, which are associated with biomechanical changes.

Importantly, this study highlighted the temperature sensitivity of the given method. Thus, for biomechanical analyses to be meaningful, knowledge of the temperature curve at the crime scene throughout the past hours to days is crucial. Temperature is a common confounder for forensic time since death estimation and also applies to the established nomogram method of Henssge [26]. Measuring both the ambient temperature and the rectal temperature of the deceased is a standard procedure in every crime scene investigation. Furthermore, local meteorological services can assist in supplying the necessary temperature data to cover the gap between the presumed time of death and the data acquired during the forensic investigation at the scene. When comparing the properties at 4 and 20 °C, the temperature sensitivity was brain region-dependent. It ranged from significant differences throughout all the four testing days in the CSmod values of the ADB samples to no significant differences for any of the tested properties in M&P samples. Especially considering the fair diagnostic ability of M&P samples to determine a PMI of at least three days at 4 °C, its temperature stability between 4 and 20 °C makes it a promising sampling region for biomechanical time since death analyses.

The potential use of biomechanical testing for forensic time since death estimation was demonstrated on ovine tissues. As a next step, the findings should be verified on human tissues. So far, the method has been explored at 4 and 20 °C, representing relevant temperatures from a practical perspective. As it seems to be impractical to test each and every temperature value, interpolation might be an option to fill the gaps after some other relevant temperature values have been investigated, e.g. around the human body temperature. Moreover, different body tissues should be explored applying the here stated protocol as their varying tissue structure and location will cause them to biomechanically respond at different times post-mortem, thereby extending the use of biomechanical analyses for time since death estimations beyond the data stated for brain tissue.

Firstly, this study had limitations in sample population size. However, increasing the population size is unlikely to improve ROC values, which are the major tool for assessing classification potential. Secondly, this study was conducted on ovine tissues. Although the findings provide a thorough starting point for analyses on human samples, the cut-off values should be verified on human tissues to be valid for forensic purposes. Thirdly, differences in brain size and concomitantly in sampling regions may have contributed to the standard deviations of the presented biomechanical properties, as sample harvesting may have slightly varied each time. Fourthly, it was initially intended to use brains only from healthy sheep in this study. However, as the brains were not tested for medical conditions, unknown (veterinary) pathological alterations invisible upon gross inspection might have influenced the investigated biomechanical properties. Fifthly, handling times within the 24-hour intervals might have varied by minutes. Sixthly, the PBS solution used for sample storage between retrieval and testing might have impacted the results, for example, through tissue swelling.