Recent Advances in the Molecular Imaging of Programmed Cell Death: Part II—Non–Probe-Based MRI, Ultrasound, and Optical Clinical Imaging Techniques

A review of the application of machine learning in molecular imaging

Molecular imaging (MI) is a science that uses imaging methods to reflect the changes of molecular level in living state and conduct qualitative and quantitative studies on its biological behaviors in imaging. Optical molecular imaging (OMI) and nuclear medical imaging are two key research fields of MI. OMI technology refers to the optical information generated by the imaging target (such as tumors) due to drug intervention and other reasons. By collecting the optical information, researchers can track the motion trajectory of the imaging target at the molecular level. Owing to its high specificity and sensitivity, OMI has been widely used in preclinical research and clinical surgery. Nuclear medical imaging mainly detects ionizing radiation emitted by radioactive substances. It can provide molecular information for early diagnosis, effective treatment and basic research of diseases, which has become one of the frontiers and hot topics in the field of medicine in the world today. Both OMI and nuclear medical imaging technology require a lot of data processing and analysis. In recent years, artificial intelligence technology, especially neural network-based machine learning (ML) technology, has been widely used in MI because of its powerful data processing capability. It provides a feasible strategy to deal with large and complex data for the requirement of MI. In this review, we will focus on the applications of ML methods in OMI and nuclear medical imaging.

Personalized radiation dosimetry for PRRT-how many scans are really required?

Purpose

Over recent years, peptide receptor radiotherapy (PRRT) has been recognized as an effective treatment for patients with metastatic neuroendocrine tumors (NETs). Personalized dosimetry can contribute to improve the outcome of peptide receptor radiotherapy (PRRT) in patients with metastatic NETs. Dosimetry can aid treatment planning, ensuring that absorbed dose to vulnerable normal organs (kidneys and bone marrow) does not exceed safe limits over serial treatments, and that absorbed dose to tumor is sufficient. Absorbed dose is estimated from a series of post-treatment SPECT/CT images. Total self-dose is proportional to the integral under the time activity concentration curve (TACC). Method dependence of image-based absorbed dose calculations has been previously investigated, and we set out here to extend previous work by examining implications of number of data points in the TACC and the numerical integration methods used in estimating absorbed dose.

Methods

In this retrospective study, absorbed dose estimates and effective half-lives were calculated by fitting curves to TACCs for normal organs and tumors in 30 consecutive patients who underwent a series of 4 post-treatment SPECT/CT scans at 4 h, 24 h, 4–5 days, and 1 week following ¹⁷⁷Lu-DOTATATE PRRT. We examined the effects of including only 2 or 3 rather than all 4 data points in the TACC, and the effect of numerical integration method (mono-exponential alone or in combination with trapezoidal rule) on the absorbed dose and half-life estimates. Our current method is the combination of trapezoidal rule over the first 24 h, with mono-exponential fit thereafter extrapolated to infinity. The other methods were compared to this current method.

Results

Differences in absorbed dose and effective half-life between the current method and estimates based only on the second, third, and fourth scans were very small (mean differences < 2.5%), whereas differences between the current method and 4-point mono-exponential fit were higher (mean differences < 5%) with a larger range. It appears that in a 4-point mono-exponential fit the early (4 h) time point may skew results, causing some large errors. Differences between the current method and values based on only 2 time points were relatively small (mean differences < 3.5%) when the 24 h and 1 week scans were used, but when the 24 h and 4–5 days scans, or the 4–5 days and 1 week scans were used, differences were greater.

Conclusion

This study indicates that for ¹⁷⁷Lu-DOTATATE PRRT, accurate estimates of absorbed dose for organs and tumors may be estimated from scans at 24 h, 72 h, and 1 week post-treatment without an earlier scan. It may even be possible to cut out the 72 h scan, though the uncertainty increases. However, further work on more patients is required to validate this.

Recent methodology advances in fluorescence molecular tomography

Molecular imaging (MI) is a novel imaging discipline that has been continuously developed in recent years. It combines biochemistry, multimodal imaging, biomathematics, bioinformatics, cell & molecular physiology, biophysics, and pharmacology, and it provides a new technology platform for the early diagnosis and quantitative analysis of diseases, treatment monitoring and evaluation, and the development of comprehensive physiology. Fluorescence Molecular Tomography (FMT) is a type of optical imaging modality in MI that captures the three-dimensional distribution of fluorescence within a biological tissue generated by a specific molecule of fluorescent material within a biological tissue. Compared with other optical molecular imaging methods, FMT has the characteristics of high sensitivity, low cost, and safety and reliability. It has become the research frontier and research hotspot of optical molecular imaging technology. This paper took an overview of the recent methodology advances in FMT, mainly focused on the photon propagation model of FMT based on the radiative transfer equation (RTE), and the reconstruction problem solution consist of forward problem and inverse problem. We introduce the detailed technologies utilized in reconstruction of FMT. Finally, the challenges in FMT were discussed. This survey aims at summarizing current research hotspots in methodology of FMT, from which future research may benefit.

Impact of motion compensation and partial volume correction for 18F-NaF PET/CT imaging of coronary plaque

Recent studies have suggested that ¹⁸F-NaF-PET enables visualization and quantification of plaque micro-calcification in the coronary tree. However, PET imaging of plaque calcification in the coronary arteries is challenging because of the respiratory and cardiac motion as well as partial volume effects. The objective of this work is to implement an image reconstruction framework, which incorporates compensation for respiratory as well as cardiac motion (MoCo) and partial volume correction (PVC), for cardiac ¹⁸F-NaF PET imaging in PET/CT. We evaluated the effect of MoCo and PVC on the quantification of vulnerable plaques in the coronary arteries. Realistic simulations (Biograph TPTV, Biograph mCT) and phantom acquisitions (Biograph mCT) were used for these evaluations. Different uptake values in the calcified plaques were evaluated in the simulations, while three 'plaque-type' lesions of 36, 31 and 18 mm³ were included in the phantom experiments. After validation, the MoCo and PVC methods were applied in four pilot NaF-PET patient studies. In all cases, the MoCo-based image reconstruction was performed using the STIR software. The PVC was obtained from a local projection (LP) method, previously evaluated in preclinical and clinical PET. The results obtained show a significant increase of the measured lesion-to-background ratios (LBR) in the MoCo  +  PVC images. These ratios were further enhanced when using directly the tissue-activities from the LP method, making this approach more suitable for the quantitative evaluation of coronary plaques. When using the LP method on the MoCo images, LBR increased between 200% and 1119% in the simulated data, between 212% and 614% in the phantom experiments and between 46% and 373% in the plaques with positive uptake observed in the pilot patients. In conclusion, we have built and validated a STIR framework incorporating MoCo and PVC for ¹⁸F-NaF PET imaging of coronary plaques. First results indicate an improved quantification of plaque-type lesions.

Optical Imaging Paves the Way for Autophagy Research

Autophagy is a degradation process in eukaryotic cells that recycles cellular components for nutrition supply under environmental stress and plays a double-edged role in development of major human diseases. Noninvasive optical imaging enables us to clearly visualize various classes of structures involved in autophagy at macroscopic and microscopic dynamic levels. In this review, we discuss important trends of emerging optical imaging technologies used to explore autophagy and provide insights into the mechanistic investigation and structural study of autophagy in mammalian cells. Some exciting new prospects and future research directions regarding optical imaging techniques in this field are also highlighted.

Effects of body habitus on internal radiation dose calculations using the 5-year-old anthropomorphic male models

Computational phantoms are commonly used in internal radiation dosimetry to assess the amount and distribution pattern of energy deposited in various parts of the human body from different internal radiation sources. Radiation dose assessments are commonly performed on predetermined reference computational phantoms while the argument for individualized patient-specific radiation dosimetry exists. This study aims to evaluate the influence of body habitus on internal dosimetry and to quantify the uncertainties in dose estimation correlated with the use of fixed reference models. The 5-year-old IT'IS male phantom was modified to match target anthropometric parameters, including body weight, body height and sitting height/stature ratio (SSR), determined from reference databases, thus enabling the creation of 125 5-year-old habitus-dependent male phantoms with 10th, 25th, 50th, 75th and 90th percentile body morphometries. We evaluated the absorbed fractions and the mean absorbed dose to the target region per unit cumulative activity in the source region (S-values) of F-18 in 46 source regions for the generated 125 anthropomorphic 5-year-old hybrid male phantoms using the Monte Carlo N-Particle eXtended general purpose Monte Carlo transport code and calculated the absorbed dose and effective dose of five ¹⁸F-labelled radiotracers for children of various habitus. For most organs, the S-value of F-18 presents stronger statistical correlations with body weight, standing height and sitting height than BMI and SSR. The self-absorbed fraction and self-absorbed S-values of F-18 and the absorbed dose and effective dose of ¹⁸F-labelled radiotracers present with the strongest statistical correlations with body weight. For ¹⁸F-Amino acids, ¹⁸F-Brain receptor substances, ¹⁸F-FDG, ¹⁸F-L-DOPA and ¹⁸F-FBPA, the mean absolute effective dose differences between phantoms of different habitus and fixed reference models are 11.4%, 11.3%, 10.8%, 13.3% and 11.4%, respectively. Total body weight, standing height and sitting height have considerable effects on human internal dosimetry. Radiation dose calculations for individual subjects using the most closely matched habitus-dependent computational phantom should be considered as an alternative to improve the accuracy of the estimates.

Compactly Supported Radial Basis Function-Based Meshless Method for Photon Propagation Model of Fluorescence Molecular Tomography

Fluorescence Molecular Tomography (FMT) is a powerful imaging modality for the research of cancer diagnosis, disease treatment and drug discovery. Via three-dimensional (3-D) imaging reconstruction, it can quantitatively and noninvasively obtain the distribution of fluorescent probes in biological tissues. Currently, photon propagation of FMT is conventionally described by the Finite Element Method (FEM), and it can obtain acceptable image quality. However, there are still some inherent inadequacies in FEM, such as time consuming, discretization error and inflexibility in mesh generation, which partly limit its imaging accuracy. To further improve the solving accuracy of photon propagation model (PPM), we propose a novel compactly supported radial basis functions (CSRBFs)-based meshless method (MM) to implement the PPM of FMT. We introduced a series of independent nodes and continuous CSRBFs to interpolate the PPM, which can avoid complicated mesh generation. To analyze the performance of the proposed MM, we carried out numerical heterogeneous mouse simulation to validate the simulated surface fluorescent measurement. Then we performed an in vivo experiment to observe the tomographic reconstruction. The experimental results confirmed that our proposed MM could obtain more similar surface fluorescence measurement with the golden standard (Monte-Carlo method), and more accurate reconstruction result was achieved via MM in in vivo application.

ASSESSMENT OF EFFECTIVE DOSE FROM CONE BEAM CT IMAGING IN SPECT/CT EXAMINATION IN COMPARISON WITH OTHER MODALITIES

The aim of this study was to assess radiation dose from the cone beam computed tomography (CBCT) component of single photon emission tomography/computed tomography (SPECT/CT) examinations and to compare it with the radiopharmaceutical related dose as well as dose from multidetector computed tomography (MDCT). Effective dose (ED) from computed tomography (CT) was estimated using dose-length product values and anatomy-specific conversion factors. The contribution from the SPECT component was evaluated using ED per unit administered activity for the radiopharmaceuticals listed in the International Commission on Radiological Protection Publications 80 and 106. With the exception of cardiac studies (0.11 mSv), the CBCT dose (3.96-6.04 mSv) was similar to that from the radiopharmaceutical accounting for 29-56 % of the total ED from the examination. In comparison with MDCT examinations, the CBCT dose was 48 and 42 % lower for abdomen/pelvis and chest/abdomen/pelvis scans, respectively, while in the chest the CBCT scan resulted in higher dose (23 %). Radiation dose from the CT component should be taken into consideration when evaluating total SPECT/CT patient dose.

Assesment of apoptosis induced changes in scattering using optical coherence tomography

The aim of this study is to identify changes in scattering with optical coherence tomography (OCT) and relate these measurements with mitochondrial changes during the initiation of apoptosis. Human retinal pigment epithelial cells were cultured and apoptosis was induced using 10% alcohol. Using the attenuation coefficient and backscattering, changes were measured during cell death in a cell-pellet and monolayer respectively. To confirm apoptosis, fluorescent activated cell sorting was used. Mitochondrial activity during apoptosis was assessed using an oxidative stress assay and fluorescent confocal microscopy. Pelleted apoptotic cells measured with OCT showed a clear rise while untreated cells showed a very small increase in attenuation coefficient. Monolayered apoptotic cells displayed a distinct increase, while untreated cells showed a small increase in the backscattering. Apoptosis was confirmed by FACS experiments. Mitochondrial changes during the onset of apoptosis were also measured. The results demonstrate that apoptotic cell death could be monitored in real-time by OCT. Changes in the scattering after induction of apoptosis are likely to be related to changes in the intracellular morphology. Oxidative stress-induced mitochondrial swelling could be responsible for the initial increase, while cell blebbing and secondary necrosis subsequently for the observed decrease in scattering.

Detection of cancer cells using SapC-DOPS nanovesicles

Unlike normal cells, cancer cells express high levels of phosphatidylserine on the extracellular leaflet of their cell membrane. Exploiting this characteristic, our lab developed a therapeutic agent that consists of the fusogenic protein, saposin C (SapC) which is embedded in dioleoylphosphatidylserine (DOPS) vesicles. These nanovesicles selectively target cancer cells and induce apoptosis. Here we review the data supporting use of SapC-DOPS to locate tumors for surgical resection or for treatment. In addition, there is important evidence suggesting that SapC-DOPS may also prove to be an effective novel cancer therapeutic reagent. Given that SapC-DOPS is easily labeled with lipophilic dyes, it has been combined with the far-red fluorescent dye, CellVue Maroon (CVM), for tumor targeting studies. We also have used contrast agents incorporated in the SapC-DOPS nanovesicles for computed tomography and magnetic resonance imaging, and review that data here. Administered intravenously, the fluorescently labeled SapC-DOPS traversed the blood–brain tumor barrier enabling identification of brain tumors. SapC-DOPS-CVM also detected a variety of other mouse tumors in vivo, rendering them observable by optical imaging using IVIS and multi-angle rotational optical imaging. Dye is detected within 30 min and remains within tumor for at least 7 days, whereas non-tumor tissues were unstained (some dye observed in the liver was transient, likely representing degradation products). Additionally, labeled SapC-DOPS ex vivo delineated tumors in human histological specimens. SapC-DOPS can also be labeled with contrast reagents for computed tomography or magnetic resonance imaging. In conclusion, labeled SapC-DOPS provides a convenient, specific, and nontoxic method for detecting tumors while concurrently offering a therapeutic benefit.

Meshless reconstruction method for fluorescence molecular tomography based on compactly supported radial basis function

Fluorescence molecular tomography (FMT) is a promising tool in the study of cancer, drug discovery, and disease diagnosis, enabling noninvasive and quantitative imaging of the biodistribution of fluorophores in deep tissues via image reconstruction techniques. Conventional reconstruction methods based on the finite-element method (FEM) have achieved acceptable stability and efficiency. However, some inherent shortcomings in FEM meshes, such as time consumption in mesh generation and a large discretization error, limit further biomedical application. In this paper, we propose a meshless method for reconstruction of FMT (MM-FMT) using compactly supported radial basis functions (CSRBFs). With CSRBFs, the image domain can be accurately expressed by continuous CSRBFs, avoiding the discretization error to a certain degree. After direct collocation with CSRBFs, the conventional optimization techniques, including Tikhonov, L1-norm iteration shrinkage (L1-IS), and sparsity adaptive matching pursuit, were adopted to solve the meshless reconstruction. To evaluate the performance of the proposed MM-FMT, we performed numerical heterogeneous mouse experiments and in vivo bead-implanted mouse experiments. The results suggest that the proposed MM-FMT method can reduce the position error of the reconstruction result to smaller than 0.4 mm for the double-source case, which is a significant improvement for FMT.

Design of a functional cyclic HSV1-TK reporter and its application to PET imaging of apoptosis

Positron emission tomography (PET) is a sensitive and noninvasive imaging method that is widely used to explore molecular events in living subjects. PET can precisely and quantitatively evaluate cellular apoptosis, which has a crucial role in various physiological and pathological processes. In this protocol, we describe the design and use of an engineered cyclic herpes simplex virus 1-thymidine kinase (HSV1-TK) PET reporter whose kinase activity is specifically switched on by apoptosis. The expression of cyclic TK (cTK) in healthy cells leads to inactive product, whereas the activation of apoptosis through the caspase-3 pathway cleaves cTK, thus restoring its activity and enabling PET imaging. In addition to detailing the design and construction of the cTK plasmid in this protocol, we include assays for evaluating the function and specificity of the cTK reporter in apoptotic cells, such as assays for measuring the cell uptake of PET tracer in apoptotic cells, correlating doxorubicin (Dox)-induced cell apoptosis to cTK function recovery, and in vivo PET imaging of cancer cell apoptosis, and we also include corresponding data acquisition methods. The time to build the entire cTK reporter is ∼2-3 weeks. The selection of a stable cancer cell line takes ∼4-6 weeks. The time to implement assays regarding cTK function in apoptotic cells and the in vivo imaging varies depending on the experiment. The cyclization strategy described in this protocol can also be adapted to create other reporter systems for broad biomedical applications.

A Novel Region Reconstruction Method for Fluorescence Molecular Tomography

Fluorescence molecular tomography (FMT) could exploit the distribution of fluorescent biomarkers that target tumors accurately and effectively, which enables noninvasive real-time 3-D visualization as well as quantitative analysis of small tumors in small animal studies in vivo. Due to the difficulties of reconstruction, continuous efforts are being made to find more practical and efficient approaches to accurately obtain the characteristics of fluorescent regions inside biological tissues. In this paper, we propose a region reconstruction method for FMT, which is defined as an L1-norm regularization piecewise constant level set approach. The proposed approach adopts a priori information including the sparsity of the fluorescent sources and the fluorescent contrast between the target and background. When the contrast of different fluorescent sources is low to a certain degree, our approach can simultaneously solve the detection and characterization problems for the reconstruction of FMT. To evaluate the performance of the region reconstruction method, numerical phantom experiments and in vivo bead-implanted mouse experiments were performed. The results suggested that the proposed region reconstruction method was able to reconstruct the features of the fluorescent regions accurately and effectively, and the proposed method was able to be feasibly adopted in in vivo application.

Reconstruction of fluorescence molecular tomography via a nonmonotone spectral projected gradient pursuit method

Fluorescence molecular tomography (FMT) is a promising imaging technique in preclinical research, enabling three-dimensional location of the specific tumor position for small animal imaging. However, FMT presents a challenging inverse problem that is quite ill-posed and ill-conditioned. Thus, the reconstruction of FMT faces various challenges in its robustness and efficiency. We present an FMT reconstruction method based on nonmonotone spectral projected gradient pursuit (NSPGP) with /₁-norm optimization. At each iteration, a spectral gradient-projection method approximately minimizes a least-squares problem with an explicit one-norm constraint. A nonmonotone line search strategy is utilized to get the appropriate updating direction, which guarantees global convergence. Additionally, the Barzilai-Borwein step length is applied to build the optimal step length, further improving the convergence speed of the proposed method. Several numerical simulation studies, including multisource cases as well as comparative analyses, have been performed to evaluate the performance of the proposed method. The results indicate that the proposed NSPGP method is able to ensure the accuracy, robustness, and efficiency of FMT reconstruction. Furthermore, an in vivo experiment based on a heterogeneous mouse model was conducted, and the results demonstrated that the proposed method held the potential for practical applications of FMT.

EGF receptor-targeting peptide conjugate incorporating a near-IR fluorescent dye and a novel 1,4,7-triazacyclononane-based (64)Cu(II) chelator assembled via click chemistry

A new Boc-protected 1,4,7-triazacyclononane (TACN)-based pro-chelator compound featuring a "clickable" azidomethylpyridine pendant has been developed as a building block for the construction of multimodal imaging agents. Conjugation to a model alkyne (propargyl alcohol), followed by deprotection, generates a pentadentate ligand, as confirmed by X-ray crystallographic analysis of the corresponding distorted square-pyramidal Cu(II) complex. The ligand exhibits rapid (64)Cu(II)-binding kinetics (>95% radiochemical yield in <5 min) and a high resistance to demetalation. It may thus prove suitable for use in (64)Cu(II)-based in vivo positron emission tomography (PET). The new chelating building block has been applied to the construction of a bimodal (PET/fluorescence) peptide-based imaging probe targeting the epidermal growth factor (EGF) receptor, which is highly overexpressed on the surface of several types of cancer cells. The probe consists of a hexapeptide sequence, Leu-Ala-Arg-Leu-Leu-Thr (designated "D4"), followed by a Cys-β-Ala-β-Ala spacer, then a β-homopropargylglycine residue with the TACN-based chelator "clicked" to its side chain. A sulfonated near-infrared (NIR) fluorescent cyanine dye (sulfo-Cy5) was introduced at the N-terminus to study the EGF receptor-binding ability of the probe by laser-fluorescence spectroscopy. Binding was also confirmed by coimmunoprecipitation methods, and an apparent dissociation constant (Kd) of ca. 10 nM was determined from radioactivity-based measurements of probe binding to two EGF receptor-expressing cell lines (FaDu and A431). The probe is shown to be a biased or partial allosteric agonist of the EGF receptor, inducing phosphorylation of Thr669 and Tyr992, but not the Tyr845, Tyr998, Tyr1045, Tyr1068, or Tyr1148 residues of the receptor, in the absence of the orthosteric EGF ligand. Additionally, the probe was found to suppress the EGF-stimulated autophosphorylation of these latter residues, indicating that it is also a noncompetitive antagonist.

A cyclic HSV1-TK reporter for real-time PET imaging of apoptosis

The coordination of cell proliferation and programmed death (apoptosis) is essential for normal physiology, and imbalance in these two opposing processes is implicated in various diseases. Objective and quantitative noninvasive imaging of apoptosis would significantly facilitate rapid screening as well as validation of therapeutic chemicals. Herein, we molecularly engineered an apoptosis switch-on PET-based cyclic herpes simplex virus type 1-thymidine kinase reporter (cTK266) containing a caspase-3 recognition domain as the switch. Translation of the reporter and protein splicing in healthy mammalian cells produce an inactive cyclic chimera. Upon apoptosis, caspase-3-specific cleavage of the circular product occurs, resulting in the restoration of the thymidine kinase activity, which can be detected in living cells and animals by noninvasive PET imaging. Our results showed the high sensitivity of this reporter in dynamic and quantitative imaging of apoptosis in living subjects. This reporter could be applied as a valuable tool for high-throughput functional screening of proapoptotic and antiapoptotic compounds in preclinical models in drug development, and monitoring the destination of therapeutic cells in clinical settings.

Evaluation of radiation dose to anthropomorphic paediatric models from positron-emitting labelled tracers

PET uses specific molecules labelled with positron-emitting radionuclides to provide valuable biochemical and physiological information. However, the administration of radiotracers to patients exposes them to low-dose ionizing radiation, which is a concern in the paediatric population since children are at a higher cancer risk from radiation exposure than adults. Therefore, radiation dosimety calculations for commonly used positron-emitting radiotracers in the paediatric population are highly desired. We evaluate the absorbed dose and effective dose for 19 positron-emitting labelled radiotracers in anthropomorphic paediatric models including the newborn, 1-, 5-, 10- and 15-year-old male and female. This is achieved using pre-calculated S-values of positron-emitting radionuclides of UF-NCI paediatric phantoms and published biokinetic data for various radiotracers. The influence of the type of anthropomorphic model, tissue weight factors and direct human- versus mouse-derived biokinetic data on the effective dose for paediatric phantoms was also evaluated. In the case of (18)F-FDG, dosimetry calculations of reference paediatric patients from various dose regimens were also calculated. Among the considered radiotracers, (18)F-FBPA and (15)O-water resulted in the highest and lowest effective dose in the paediatric phantoms, respectively. The ICRP 103 updated tissue-weighting factors decrease the effective dose in most cases. Substantial differences of radiation dose were observed between direct human- versus mouse-derived biokinetic data. Moreover, the effect of using voxel- versus MIRD-type models on the calculation of the effective dose was also studied. The generated database of absorbed organ dose and effective dose for various positron-emitting labelled radiotracers using new generation computational models and the new ICRP tissue-weighting factors can be used for the assessment of radiation risks to paediatric patients in clinical practice. This work also contributes to a better understanding of the factors influencing patient-specific radiation dose calculation.

The promise and limits of PET texture analysis

Metabolic heterogeneity is a recognized characteristic of malignant tumors. Positron emission tomography (PET) texture analysis evaluated intratumoral heterogeneity in the uptake of (18)F-fluorodeoxyglucose. There were recent evidences that PET textural features were of prognostic significance in patients with different solid tumors. Unfortunately, there are still crucial standardization challenges to transform PET texture parameters from their current use as research tools into the arena of validated technologies for use in oncology practice. Testing its generalizability, robustness, consistency, and limitations is necessary before implementing it in daily patient care.