Development and validation of a radiomic model for the diagnosis of dopaminergic denervation on [18F]FDOPA PET/CT

Purpose

FDOPA PET shows good performance for the diagnosis of striatal dopaminergic denervation, making it a valuable tool for the differential diagnosis of Parkinsonism. Textural features are image biomarkers that could potentially improve the early diagnosis and monitoring of neurodegenerative parkinsonian syndromes. We explored the performances of textural features for binary classification of FDOPA scans.

Methods

We used two FDOPA PET datasets: 443 scans for feature selection, and 100 scans from a different PET/CT system for model testing. Scans were labelled according to expert interpretation (dopaminergic denervation versus no dopaminergic denervation). We built LASSO logistic regression models using 43 biomarkers including 32 textural features. Clinical data were also collected using a shortened UPDRS scale.

Results

The model built from the clinical data alone had a mean area under the receiver operating characteristics (AUROC) of 63.91. Conventional imaging features reached a maximum score of 93.47 but the addition of textural features significantly improved the AUROC to 95.73 (p < 0.001), and 96.10 (p < 0.001) when limiting the model to the top three features: GLCM_Correlation, Skewness and Compacity. Testing the model on the external dataset yielded an AUROC of 96.00, with 95% sensitivity and 97% specificity. GLCM_Correlation was one of the most independent features on correlation analysis, and systematically had the heaviest weight in the classification model.

Conclusion

A simple model with three radiomic features can identify pathologic FDOPA PET scans with excellent sensitivity and specificity. Textural features show promise for the diagnosis of parkinsonian syndromes.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00259-022-05816-7.

Gut CD4+ T cell phenotypes are a continuum molded by microbes, not by TH archetypes

CD4+ effector lymphocytes (Teff) are traditionally classified by the cytokines they produce. To determine the states that Teff cells actually adopt in frontline tissues in vivo, we applied single-cell transcriptome and chromatin analyses to colonic Teff cells in germ-free or conventional mice or in mice after challenge with a range of phenotypically biasing microbes. Unexpected subsets were marked by the expression of the interferon (IFN) signature or myeloid-specific transcripts, but transcriptome or chromatin structure could not resolve discrete clusters fitting classic helper T cell (TH) subsets. At baseline or at different times of infection, transcripts encoding cytokines or proteins commonly used as TH markers were distributed in a polarized continuum, which was functionally validated. Clones derived from single progenitors gave rise to both IFN-γ- and interleukin (IL)-17-producing cells. Most of the transcriptional variance was tied to the infecting agent, independent of the cytokines produced, and chromatin variance primarily reflected activities of activator protein (AP)-1 and IFN-regulatory factor (IRF) transcription factor (TF) families, not the canonical subset master regulators T-bet, GATA3 or RORγ.

Deep learning of immune cell differentiation

Although we know many sequence-specific transcription factors (TFs), how the DNA sequence of cis-regulatory elements is decoded and orchestrated on the genome scale to determine immune cell differentiation is beyond our grasp. Leveraging a granular atlas of chromatin accessibility across 81 immune cell types, we asked if a convolutional neural network (CNN) could learn to infer cell type-specific chromatin accessibility solely from regulatory DNA sequences. With a tailored architecture and an ensemble approach to CNN parameter interpretation, we show that our trained network ("AI-TAC") does so by rediscovering ab initio the binding motifs for known regulators and some unknown ones. Motifs whose importance is learned virtually as functionally important overlap strikingly well with positions determined by chromatin immunoprecipitation for several TFs. AI-TAC establishes a hierarchy of TFs and their interactions that drives lineage specification and also identifies stage-specific interactions, like Pax5/Ebf1 vs. Pax5/Prdm1, or the role of different NF-κB dimers in different cell types. AI-TAC assigns Spi1/Cebp and Pax5/Ebf1 as the drivers necessary for myeloid and B lineage fates, respectively, but no factors seemed as dominantly required for T cell differentiation, which may represent a fall-back pathway. Mouse-trained AI-TAC can parse human DNA, revealing a strikingly similar ranking of influential TFs and providing additional support that AI-TAC is a generalizable regulatory sequence decoder. Thus, deep learning can reveal the regulatory syntax predictive of the full differentiative complexity of the immune system.

Vacuum-ultraviolet frequency-modulation spectroscopy

Frequency-modulation (FM) spectroscopy has been extended to the vacuum-ultraviolet (VUV) range of the electromagnetic spectrum. Coherent VUV laser radiation is produced by resonance-enhanced sum-frequency mixing (ν_VUV=2ν_UV+ν₂) in Kr and Xe using two near-Fourier-transform-limited laser pulses of frequencies ν_UV and ν₂. Sidebands generated in the output of the second laser (ν₂) using an electro-optical modulator operating at the frequency ν_mod are directly transferred to the VUV and used to record FM spectra. Demodulation is demonstrated both at ν_mod and 2ν_mod. The main advantages of the method compared to VUV absorption spectroscopy are its background-free nature, the fact is that its implementation using table-top laser equipment is straightforward and that it can be used to record VUV absorption spectra of cold samples in skimmed supersonic beams simultaneously with laser-induced-fluorescence and photoionization spectra. To illustrate these advantages, we present VUV FM spectra of Ar, Kr, and N₂ in selected regions between 105000 cm^-1 and 122000 cm^-1.

Surface-electrode Rydberg-Stark decelerator

Hydrogen atoms in Rydberg states with principal quantum numbers between 23 and 70 have been accelerated, decelerated, and electrostatically trapped using a surface-electrode Rydberg-Stark decelerator. By applying a set of oscillating electrical potentials to a two-dimensional array of electrodes on a printed circuit board (PCB), a continuously moving, three-dimensional electric trap with a predefined velocity and acceleration is generated. From an initial longitudinal velocity of 760  m/s, final velocities of the Rydberg atoms ranging from 1200  m/s to zero velocity in the laboratory-fixed frame of reference were achieved. Accelerated or decelerated atoms were detected directly by pulsed electric-field ionization. Atoms trapped at zero mean velocity above the PCB were reaccelerated off the PCB before field ionization.

Collisional and radiative processes in adiabatic deceleration, deflection, and off-axis trapping of a Rydberg atom beam

A supersonic beam of Rydberg hydrogen atoms has been adiabatically deflected by 90°, decelerated to zero velocity in less than 25  μs, and loaded into an electric trap. The deflection has allowed the suppression of collisions with atoms in the trailing part of the gas pulse. The processes leading to trap losses, i.e., fluorescence to the ground state, and transitions and ionization induced by blackbody radiation have been monitored over several milliseconds and quantitatively analyzed.

Magnetic trapping of hydrogen after multistage Zeeman deceleration

We report the first experimental realization of magnetic trapping of a sample of cold radicals following multistage Zeeman deceleration of a pulsed supersonic beam. H atoms seeded in a supersonic expansion of Kr have been decelerated from an initial velocity of 520 m/s to 100 m/s in a 12-stage Zeeman decelerator and loaded into a magnetic quadrupole trap by rapidly switching the fields of the trap solenoids.

Radio diagnosis using 125I-fibrinogen and 203Hg-thymidine

A comparative study of 125I-fibrinogen and 203Hg-thymidine using 144 Ehrlich carcinoma-bearing mice, indicates that tumor affinity of carrier substances with covalent labeling could be better used for scintigraphic tumor diagnosis. Halogenated or possibly metallo-organic compounds offer an alternative to the metal complexes in general use up to now.