Estimation of Gestational Age from Tooth Germs: Biometric Study of DentaScan Images

Abstract: The few available studies on fetal age estimation concern very small samples, and statistical analysis is sometimes inadequate. In this survey, we used germs of deciduous teeth to estimate fetal age. Forty‐nine fetuses and 40 mandibles were scanned, and observations and measurements were made on DentaScan images. After checking their repeatability and reproducibility (analysis of variance), we defined thresholds using Fisher’s linear discriminant analysis to calculate the probability that a fetus was over or below a predefined age threshold. The forensic threshold which is of particular interest in France is 22 weeks amenorrhea. Relationships between fetal age and deciduous germ measurements were then sought by multiple linear regression. The thresholds gave very good results: 91.84% of good probability for the threshold of 22 weeks amenorrhea with no chance of error. The most precise age evaluation obtained nevertheless gave a range of ±4.6 weeks amenorrhea, so greater accuracy is still needed.     

Risk evaluation for spoofing against a sensor supplied with liveness detection

The aim of this paper is to evaluate the risks associated with the use of fake fingerprints on a livescan supplied with a method of liveness detection. The method is based on optical properties of the skin. The sensor uses several polarizations and illuminations to capture the information of the different layers of the human skin. These experiments also allow for the determination under which conditions the system is deceived and if there is an influence respectively of the nature of the fake, the mould used for the production or the individuals involved in the attack.These experiments showed that current multispectral sensors can be deceived by the use of fake fingerprints created with or without the cooperation of the subject. Fakes created from direct casts perform better than those produced by fakes created from indirect casts. The results showed that the success of the attack is influenced by two main factors. The first is the quality of the fakes, and by extension the quality of the original fingerprint. The second is the combination of the general patterns involved in the attacks since an appropriate combination can strongly increase the rates of successful attacks.     

Detection of amitriptyline, nortriptyline and bromazepam in liver, CSF and hair in the homicidal poisoning of a one-month-old girl autopsied 8 months after death

We reported on the death by poisoning of a one-month-old baby that had followed the death of one of her sister (due to cyamemazine overdose). Exhumation of the corpse was done 8 months after burial and revealed the presence of amitriptyline. Parent drug and its metabolite were analysed by HPLC-MS/MS in positive ionisation mode on a C(18) analytical column using a gradient of acetonitrile and 2mM formate buffer at pH=3. Quantification is based on the main ion m/z=233, the common product ion of nortriptyline (MH(+), m/z 264), amitriptyline (MH(+), m/z 278) and nortriptyline D3 used as internal standard (MH(+), m/z 267). Amitriptyline and nortriptyline in the liver were measured at a concentration of 29.8 and 3.6 μg/g, respectively. Hair analyses revealed the presence of amitriptyline and nortriptyline at concentrations of 1811 and 43 pg/mg, respectively, while complementary analyses showed the presence of bromazepam in the hair at a concentration of 740 pg/mg, thus documenting previous administrations. The mother confessed later having used the drinkable form of the pharmaceutical LAROXYL(?) by pouring the content of a 20 ml bottle (at 40 mg/ml) into the feeding-bottle of her child. The milk was sweet but still bitter and following the testimony of a close relative, the whole family helped to feed the crying baby.     

Vulnerabilities of fingerprint reader to fake fingerprints attacks

The purpose of this research is to assess the vulnerabilities of a high resolution fingerprint sensor when confronted with fake fingerprints. The study has not been focused on the decision outcome of the biometric device, but essentially on the scores obtained following the comparison between a query (genuine or fake) and a template using an AFIS system. To do this, fake fingerprints of 12 subjects have been produced with and without their cooperation. These fake fingerprints have been used alongside with real fingers. The study led to three major observations: First, genuine fingerprints produced scores higher than fake fingers (translating a closer proximity) and this tendency is observed considering each subject separately. Second, scores are however not sufficient as a single measure to differentiate these samples (fake from genuine) given the variation due to the donors themselves. That explains why fingerprint readers without vitality detection can be fooled. Third, production methods and subjects greatly influence the scores obtained for fake fingerprints.