Secondary Logo

Journal Logo

Original Article

Postmortem Fetal Temperature Estimation with Magnetic Resonance Imaging: Apparent Diffusion Coefficient Measurements in the Vitreous Body and Cerebrospinal Fluid

Tijssen, Maud P.M. MD; Hofman, Paul A.M. MD, PhD; Robben, Simon G.F. MD, PhD

Author Information
Topics in Magnetic Resonance Imaging: April 2022 - Volume 31 - Issue 2 - p 25-30
doi: 10.1097/RMR.0000000000000295
  • Open


Postmortem magnetic resonance imaging (PMMR) is increasingly used. Currently, PMMR is an accepted method, adjunct to or as a replacement for autopsy, to confirm the prenatal diagnosis (in case of termination of pregnancy) or evaluate the cause of death (in case of intrauterine fetal death).1

Fetal PMMR has several limitations that can affect the quality and interpretation of the scan, including small size, field strength, and autolytic and maceration changes.1–3 Furthermore, the quality and signal intensity of magnetic resonance imaging (MRI) depend on temperature; for example, T1-weighted imaging, T2-weighted imaging, and diffusion-weighted imaging (DWI) experience temperature-dependent changes in signal intensity that differ between tissue types.4–8 Contrast between tissues is thus affected by temperature, and correct image interpretation requires knowledge of the fetus temperature of the fetus at the time of scan. Temperature decrease will influence the T1 and T2 relaxation times; with decreasing temperature the T1 values are initially shortened and then prolonged, while T2 values will show a linear decrease.5 If the temperature is known at the time of scanning, MRI parameters such as TE and TR can be adjusted to this temperature to improve the image contrast5,9 (eg, shortening TR with decreasing T1 value, shortening TE with decreasing T2 value).

In several cases, temperature evaluation using a thermometer is undesirable. Thermometer evaluation is invasive and can potentially damage the fetus, especially when severe maceration is present. This can influence autopsy results and forensic evaluation. The invasive character of temperature assessment can face resistance from parents realizing that their newborn has suffered enough. Further, in small fetuses, thermometer evaluation could be unreliable or even impossible due to the size of the thermometer. In case of retrospective PMMR evaluations, thermometer measurements are not possible. In these cases, there are alternative noninvasive methods based on MRI data. MR techniques currently used for this purpose include magnetic resonance spectroscopy (MRS), T1 relaxation time measurement, and apparent diffusion coefficient (ADC) based calculations.10,11 MRS is time consuming and not embedded in standard postmortem protocols, and in the absence of motion, diffusion coefficient-based calculations offer better precision than T1 relaxation time measurements.10 Furthermore, DWI is a routine sequence that is not time consuming, can be easily used, and is applicable in daily practice.

Kozak et al developed a formula to determine the temperature based on the ADC of cerebrospinal fluid (CSF).12 The calculation of the temperature based on the ADC is commonly used in vivo to calculate the brain temperature.13–15 In children and fetuses, this method is less reliable because of the small size of the ventricles and the effects of partial volume on ADC measurements.16 In addition, the ADC measurements are affected by intraventricular blood products, which are common in PMMR. The vitreous body is less susceptible to postmortem changes than the CSF.17,18 Furthermore, due to the size of the vitreous body, the risk of partial volume affecting the calculations is expected to be reduced. In the current study, two methods were evaluated by calculating the temperatures of small brains in lambs and human fetuses, using vitreous body ADC or CSF ADC and comparing the results with thermometer measurements.


Study Cohort

Two lamb heads (twins) were sequentially scanned at five different time points during the first 30 hours postmortem. The two lambs were born prematurely with a gestational age (GA) of 130 to 131 days (which correlates with a gestational age of 35 weeks in humans) and sacrificed within hours after birth. The lambs were raised and sacrificed for research obligations outside of our scope, for which ethical approval was obtained. Only the head is available for this purpose. After the first scan, the first lamb head was stored at room temperature (lamb head 1), and the second lamb head was stored in a refrigerator at 4°C (lamb head 2). Before each scan, the temperature was measured with a digital thermometer in the oral cavity and nasal cavity and by inserting the thermometer into the cervical cutting surface (at least 1 cm). The average of these three measurements was used for the analysis.

Additionally, 10 consecutive human fetuses with GA between 20 and 24 weeks, who received PMMR were included. Between birth and PMMR, the fetuses remained with their parents for social reasons (mourning) or were stored at 4°C. Informed consent was obtained for autopsy and PMMR (n = 9) or PMMR alone (n = 1). According to Dutch law, review by an ethical committee for research on anonymous retrospective data is not required. The postmortem interval (PMI), calculated from birth or documented death time until PMMR, was between 4 and 62 hours. Oral temperature was measured immediately before each scan using a digital thermometer.

MRI Protocol

MRI was performed at 3.0 T (Ingenia CX, Philips Healthcare, Best, The Netherlands) using a 32-channel 3.0 T head coil. Brain imaging included axial T2-sequence and coronal DWI. Axial DWI was also performed in fetuses. The DWI sequences were performed with a diffusion gradient b-value = 0 and b-value = 1000, pixel size of 1 × 1 mm, and slice thickness 1.5 mm. See Table 1 for the key pulse sequence parameters.

TABLE 1 - MRI Pulse Sequence Parameters for the Diffusion Weighted Images of the Lambs (Coronal) and Fetuses (Axial and Coronal) and T2-weighted Images of Both Lambs and Fetuses (Axial)
Lambs Fetuses
DWI (Coronal) T2 (Axial) DWI (Coronal) DWI (Axial) T2 (Axial)
Echo time (TE, ms) 105 80 105 108 100
Repetition time (TR, ms) 7192 4300 6583 9234 3186
FOV (mm) 120 × 120 120 × 101 120 × 120 120 × 120 100 × 69
Acquired resolution (mm2) 1.00 × 1.02 0.45 × 0.60 1.00 × 1.02 1.00 × 1.02 0.5 × 0.64
Slice thickness (mm) 1.5 1.5 1.5 1.5 2.0
DWI indicates diffusion-weighted imaging; MRI, magnetic resonance imaging.


The images were imported into our institution's picture archiving and communication system (Sectra IDS7 22.1). Circular ROIs were manually placed on the right eyeball (lambs and fetuses) and CSF (fetuses) on the ADC images (Fig. 1). ADC measurements in the fetuses were performed in the direction (axial or coronal) with the least artifacts. Due to collapsed and preexisting small ventricles, ADC measurements in the CSF were not possible in lambs.

Example of manual ROI placement in the vitreous body of a lamb on coronal apparent diffusion coefficient (ADC) and ROI placement in the vitreous body and cerebrospinal fluid of two fetuses on axial ADC.

Measurements were performed by one reader (MT) and repeated at intervals of at least 2 weeks. The mean ADC values of the two measurements were calculated and used for further analysis. Where necessary, anatomical correlation was performed by cross-referencing with axial T2-weighted images.

Temperature calculations were performed using the formula developed by Kozak et al12 where T is the temperature in degrees Celsius, D is the diffusion value of CSF in the ADC, and K is Kelvin:


Statistical Analysis

The measured and calculated temperatures, based on the ADC of the vitreous body (lambs and fetuses) or CSF (fetuses), were compared using the paired samples t-test and Pearson's correlation. The variance between the two repeated measurements was calculated. Statistical analyses were performed using SPSS software (IBM SPSS Statistics 25.0). Statistical significance was defined as P < 0.05.


ADC Value of the Vitreous Body Versus Temperature in Lamb Heads

The variance between the two repeated ADC measurements ranged from 0.0% to 1.4%, indicating good intraobserver reproducibility. A decrease in mean ADC values was observed over time in both lamb heads and was greater in lamb head 2 stored at 4°C than in lamb head 1 stored at room temperature. The results of the three temperature measurements are given in Table 2. The average of these three measurements was used for the analysis.

TABLE 2 - Results of Temperature Measurements in the Oral Cavity, Nasal Cavity, and Neck Measurements with Calculated Average Temperatures in Both Lambs (Stored at Room Temperature and Stored at 4°C) at the Five Different Timepoints
Lamb Head 1 (Stored at Room Temperature) Lamb Head 2 (Stored at 4°C)
Scan PMI (h) T Oral Cavity (°C) T Nasal Cavity (°C) Neck T (°C) Average (°C) PMI (h) T Oral Cavity (°C) T Nasal Cavity (°C) Neck T (°C) Average (°C)
1 2.20 28.7 28.0 25.5 27.4 1.55 32.5 30.8 27.0 30.1
2 4.25 25.3 24.6 23.0 24.3 4.00 20.5 19.6 22.0 20.7
3 6.50 23.5 23.4 23.0 23.3 6.30 15.5 14.5 16.5 15.5
4 22.50 21.8 22.0 22.2 22.0 22.30 8.0 9.9 8.1 8.7
5 28.10 22.5 22.2 22.2 22.3 27.50 7.8 9.8 7.7 8.4
PMI indicates postmortem interval.

Table 3 lists the measured and calculated temperatures. In lamb head 1, the measured temperatures varied from 22.0°C to 27.4°C, calculated temperatures varied between 20.4°C and 26.5°C, and the differences between the measured and calculated temperatures ranged from 0.1°C to 1.6°C. In lamb head 2, the measured temperatures varied from 8.4°C to 30.1°C, calculated temperatures varied from 8.7°C to 28.4°C and differences between measured and calculated temperatures ranged from 0.1°C to 1.7°C.

TABLE 3 - Measured (Average of Oral Cavity, Nasal Cavity, and Neck Measurements) and Calculated (Using Mean Apparent Diffusion Coefficient of the Vitreous Body) Temperatures in Two Lamb Heads at Five Different Postmortem Intervals
Lamb Head 1 (Stored at Room Temperature) Lamb Head 2 (Stored at 4°C)
Scan Postmortem Interval (h) Measured Temperature (°C) Calculated Temperature (°C) Postmortem Interval (h) Measured Temperature (°C) Calculated Temperature (°C)
1 2.20 27.4 26.5 1.55 30.1 28.4
2 4.25 24.3 24.2 4.00 20.7 20.8
3 6.50 23.3 23.0 6.30 15.5 15.8
4 22.50 22.0 20.4 22.30 8.7 8.9
5 28.10 22.3 21.2 27.50 8.4 8.7

The calculated and measured temperatures were strongly correlated (r = 0.997, P < 0.001; Fig. 2). A Bland-Altman plot was made to display the relationship between the two variables (Fig. 3). There was no significant difference between the measured temperature (mean = 20.3°C, SD = 7.29) and calculated temperature (mean = 19.8°C, SD = 6.75); t (9) = 1.94, P = 0.084.

Correlation between measured and calculated temperatures in the lambs using the apparent diffusion coefficient (ADC) of the vitreous body. Correlation line (solid line) with r = 0.997 (P = 0.001). The dotted line represents a 100% correlation reference line.
Bland-Altman plot of temperature assessed by measured average temperatures versus calculated temperatures with vitreous body (VB) ADC in lambs. Mean temperatures and difference in temperature are illustrated for the two lambs at all five timepoints. ADC, apparent diffusion coefficient.

ADC Value of Vitreous Body and CSF Versus Temperature in Fetuses

The variance between repeated measurements was 0.0% to 1.9% for vitreous body ADC and 0.0% to 2.6% for CSF ADC, indicating good reproducibility. The measured temperatures in the fetuses ranged from 7.2°C to 21.0°C, calculated temperatures ranged from 8.1°C to 20.0°C (vitreous body ADC), and 8.1°C to 17.2°C (ADC of the CSF) (Table 4). The differences between the measured and calculated temperatures ranged from 0.5°C to 3.0°C (vitreous body ADC) and from 0.4°C to 10.0°C (CSF ADC).

TABLE 4 - Measured and Calculated Temperatures of 10 Fetuses with Postmortem MRI Scans, Sorted by the Measured Temperature
Calculated Temperature (°C)
Gestation Age (Weeks + Days) Postmortem Interval (h) Measured Temperature (°C) Vitreous Body Cerebrospinal Fluid
21 + 6 22 7.2 8.7 8.7
23 + 0 62 9.5 8.1 7.5
21 + 3 17 11.3 10.7 9.9
22 + 6 19 11.6 11.1 11.2
20 + 3 62 12.0 12.5 8.1
22 + 0 27 14.9 14.3 10.2
20 + 5 13 16.6 15.9 17.2
21 + 6 4 16.8 14.9 7.2
23 + 3 21 19.6 16.6 15.5
23 + 0 22 21.0 20.0 11.0
MRI indicates magnetic resonance imaging.

The measured and calculated temperatures based on the ADC of the vitreous body were strongly correlated (r = 0.970, P < 0.001), and there was no significant difference in the measured temperature (mean = 14.1°C, SD = 4.46) and temperature calculated based on vitreous body ADC (mean = 13.3°C, SD = 3.75); t (9) = 1.98, P = 0.079. The correlation between the measured and calculated temperatures based on the CSF was weak and not significant (r = 0.522, P = 0.122) (Fig. 4). The difference in the measured temperature (mean = 14.1°C, SD = 4.46) and calculated temperature based on the CSF ADC (mean = 10.7°C, SD = 3.33) was significant (t (9) = 2.73, P = 0.023). Bland-Altman plots were made to illustrate the relationship between measured temperature versus calculated vitreous body (a) and CSF (b) temperatures (Fig. 5).

Correlations between measured and calculated temperatures in 10 fetuses, using the apparent diffusion coefficient of the vitreous body (closed circles, solid correlation line) or the cerebrospinal fluid (open circles, striped correlation line). ADC, apparent diffusion coefficient; CSF, cerebrospinal fluid. Vitreous body ADC: r = 0.970, P = 0.001; CSF ADC: r = 0.522, P = 0.122. The dotted line represents a 100% correlation reference line.
Bland-Altman plot of temperature assessment in fetuses with assessment of measured temperatures versus calculated temperatures with vitreous body ADC (A) and measured temperatures with calculated CSF ADC (B) in fetuses. Mean temperatures and difference in temperatures are illustrated for all 10 fetuses.


Determining postmortem fetal temperature using vitreous body ADC was easily feasible, and there was an excellent correlation between measured and calculated temperatures for both lamb heads (r = 0.997, P < 0.001) and fetuses (r = 0.970, P < 0.001). In contrast, the correlation between measured and calculated temperature using CSF ADC was poor (fetuses: r = 0.522, P = 0.122) and showed greater variance. This is probably due to the small volume of the ventricles and partial volume artifacts,16 postmortem changes in the composition of the CSF, and additional blood degradation products. Blood degradation products in combination with partial volume artifacts will lower the ADC value, resulting in an incorrect lower temperature with this method of temperature calculation (as shown in Fig. 4). The vitreous body remains much more stable in the early postmortem period17,18 and is less susceptible to partial volume artifacts, making temperature calculations using the vitreous body ADC superior in this setting. To our knowledge, this is the first study to evaluate the feasibility of vitreous body ADC measurement for temperature determination.

The results showed small differences between the measured and calculated temperatures using vitreous body ADC (0.5°C–3.0°C). It is unclear to what extent this is due to the different methods used or whether this reflects actual differences in temperature between the vitreous body and the oral cavity. However, from a practical point of view, small temperature differences are unlikely to have a significant effect on contrast and image interpretation. The assessment of temperature in postmortem MRI imaging is relevant because T1, T2, and DWI experience temperature-dependent changes in signal intensity, which can be different for various tissues, leading to changes in contrast.4–8 In the postmortem setting, scanning is typically performed at temperatures ranging from 0°C to 40°C. Zech et al developed equations to correct temperature in MRI quantification used for soft-tissue characterization.8 Using vitreous body DWI, temperatures can be quickly and noninvasively calculated, and MRI parameters can be adjusted accordingly.

Our results show a weak and insignificant correlation between the measured and CSF-based calculated temperatures. Theoretically, there could be a temperature difference between deep brain structures (eg, ventricles/CSF) and superficial structures (eg, the vitreous body and oral cavity). In the living human body, the temperature varies depending on the location.14,19 Mellergard found that temperature measurements in the deeper cerebral matter are higher than directly subcortical measurements, with an average difference of 0.5°C.19 Hirashima et al showed in living patients a difference between intraventricular temperature (37.4 [0.83]) and temperature at 2 cm depth (36.2 [0.95]).14 These temperature variations are probably due to variable metabolic activity at different locations. In a postmortem setting, metabolic activity stops and these temperature differences are not expected to exist. Therefore, we may assume that the subcortical temperature of the direct postmortem will not significantly differ from the intraventricular temperature in this setting.

However, with increasing PMI, temperature differences may arise between superficial and deep brain structures due to environmental influences. The same study by Hirashima et al mentioned preliminary data of water bath measurements where the temperature on the surface was prone to environmental temperature, but from 1 to 5 cm, the temperatures remained stable.14 In contrast, Kaliszan and Wujtewicz demonstrated that in postmortem settings, eyeball temperatures adjust and decrease quicker to ambient temperatures than core temperatures.20 Within a PMI of 3.30 hours, this difference could increase to 6.7°C.20 Yet, in our results, the CSF ADC calculated temperatures were actually lower than the vitreous body ADC. However, theoretically this gradient will develop as well when the body is moved from the refrigerator (4°C) to room temperature, where the core temperature will increase slower than superficial tissues and the calculated lower temperatures of the CSF of the temperatures than vitreous body can actually reflect lower temperatures instead of artifacts. To minimize this gradient as much as possible, DWI was the first sequence of our PMMR protocol, reducing the time from exiting the refrigerator to PMMR to a maximum of 30 minutes. Kaliszan and Wujtewicz reported an average difference between surface and core temperatures of 4.2°C in seven adult cases with PMI within 1 hour.20 Our study showed temperature differences up to 10°C. In addition, the degree of the calculated temperature difference did not correlate with the length of the PMI. Although the temperature difference between core and superficial tissues is a limitation of this study, the differences cannot be fully explained by gradient temperature adjustment and are probably at least partially due to postmortem chemical changes and partial volume artifacts of the CSF that lower the ADC value. Adult studies cannot be extrapolated to fetal studies because the diameter of an adult skull is not comparable with the diameter of the head of a fetus. Moreover, the fetal head is lacking an isolated thick bony skull and subcutaneous fat. Therefore an equilibrium of temperature between superficial and deep structures will be reached much faster in fetuses than adults.

Another limitation of this study is the small sample size. Only two lambs and 10 fetuses were used for the temperature calculations. However, even in this small population, the results were statistically significant. Confirmation of our results using a larger study population must be considered.

The applicability of retrospective temperature calculations goes beyond PMMR, and it may be interesting to evaluate them in vivo. In asphyxiated patients undergoing cooling therapy or patients with high fever, for example, a retrospective calculation of temperature could be relevant to image interpretation. However, the temperature changes in these settings are not as great as those that occur after the death. Further postmortem or in vivo investigations could be helpful to determine the impact of smaller temperature variations on image quality and interpretation.


In conclusion, vitreous body ADC-based temperature calculation seems a feasible method for determining body temperature in the setting of fetal postmortem MRI. Further studies are needed to expand the sample size and evaluate intra- and inter-reader agreement.


1. Thayyil S, Cleary JO, Sebire NJ, et al. Post-mortem examination of human fetuses: a comparison of whole-body high-field MRI at 9.4 T with conventional MRI and invasive autopsy. Lancet 2009; 374:467–475.
2. Arthurs OJ, Hutchinson JC, Sebire NJ. Current issues in postmortem imaging of perinatal and forensic childhood deaths. Forensic Sci Med Pathol 2017; 13:58–66.
3. Jawad N, Sebire NJ, Wade A, et al. Body weight lower limits of fetal postmortem MRI at 1.5 T. Ultrasound Obstet Gynecol 2016; 48:92–97.
4. Addison S, Arthurs OJ, Thayyil S. Post-mortem MRI as an alternative to non-forensic autopsy in foetuses and children: from research into clinical practice. Br J Radiol 2014; 87:20130621.
5. Kobayashi T, Isobe T, Shiotani S, et al. Postmortem magnetic resonance imaging dealing with low temperature objects. Magn Reson Med Sci 2010; 9:101–108.
6. Ruder TD, Hatch GM, Siegenthaler L, et al. The influence of body temperature on image contrast in post mortem MRI. Eur J Radiol 2012; 81:1366–1370.
7. Wagensveld IM, Blokker BM, Wielopolski PA, et al. Total-body CT and MR features of postmortem change in in-hospital deaths. PLoS One 2017; 12:e0185115.
8. Zech WD, Schwendener N, Persson A, et al. Temperature dependence of postmortem MR quantification for soft tissue discrimination. Eur Radiol 2015; 25:2381–2389.
9. Ruder TD, Thali MJ, Hatch GM. Essentials of forensic post-mortem MR imaging in adults. Br J Radiol 2014; 87:20130567.
10. Quesson B, de Zwart JA, Moonen CT. Magnetic resonance temperature imaging for guidance of thermotherapy. J Magn Reson Imaging 2000; 12:525–533.
11. Sumida K, Sato N, Ota M, et al. Intraventricular temperature measured by diffusion-weighted imaging compared with brain parenchymal temperature measured by MRS in vivo. NMR Biomed 2016; 29:890–895.
12. Kozak LR, Bango M, Szabo M, et al. Using diffusion MRI for measuring the temperature of cerebrospinal fluid within the lateral ventricles. Acta Paediatr 2010; 99:237–243.
13. Sakai K, Yamada K, Mori S, et al. Age-dependent brain temperature decline assessed by diffusion-weighted imaging thermometry. NMR Biomed 2011; 24:1063–1067.
14. Hirashima Y, Takaba M, Endo S, et al. Intracerebral temperature in patients with hydrocephalus of varying aetiology. J Neurol Neurosurg Psychiatry 1998; 64:792–794.
15. Hasan KM, Lincoln JA, Nelson FM, et al. Lateral ventricular cerebrospinal fluid diffusivity as a potential neuroimaging marker of brain temperature in multiple sclerosis: a hypothesis and implications. Magn Reson Imaging 2015; 33:262–269.
16. Wagner MW, Stern SE, Oshmyansky A, et al. The role of ADC-based thermometry in measuring brain intraventricular temperature in children. J Neuroimaging 2016; 26:315–323.
17. Chandrakanth HV, Kanchan T, Balaraj BM, et al. Postmortem vitreous chemistry—an evaluation of sodium, potassium and chloride levels in estimation of time since death (during the first 36 h after death). J Forensic Leg Med 2013; 20:211–216.
18. Belsey SL, Flanagan RJ. Postmortem biochemistry: current applications. J Forensic Leg Med 2016; 41:49–57.
19. Mellergard P. Intracerebral temperature in neurosurgical patients: intracerebral temperature gradients and relationships to consciousness level. Surg Neurol 1995; 43:91–95.
20. Kaliszan M, Wujtewicz M. Eye temperature measured after death in human bodies as an alternative method of time of death estimation in the early post mortem period. A successive study on new series of cases with exactly known time of death. Leg Med (Tokyo) 2019; 38:10–13.

autopsy; fetus; magnetic resonance imaging; postmortem changes; temperature

Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc.