Filtern
Volltext vorhanden
- ja (25)
Gehört zur Bibliographie
- ja (25)
Erscheinungsjahr
- 2023 (25) (entfernen)
Dokumenttyp
Schlagworte
- magnetic resonance imaging (3)
- CT angiography (2)
- cartilage (2)
- computed tomography (2)
- translational research (2)
- AMADEUS (1)
- Bildgebendes Verfahren (1)
- Echtzeit (1)
- Gallengangsstenose (1)
- Gefäßverschluss (1)
- Hand (1)
- Herzfunktion (1)
- Herzmuskel (1)
- Hyperämie (1)
- Interventionsradiologie (1)
- Intravenöse Applikation (1)
- KOOS (1)
- Karpus (1)
- Kernspintomografie (1)
- Kernspintomographie (1)
- Kontrastmittel (1)
- Lungenembolie (1)
- MRT (1)
- Onkologie (1)
- PEFB (1)
- PROM (1)
- SL-Band (1)
- Stent (1)
- TFCC (1)
- Trauma (1)
- Triangulärer fibrokartilaginärer Komplex (1)
- Vascular closure devices (1)
- Vena cava inferior (1)
- X-ray computed (1)
- Zangenbiopsie (1)
- access site complications (1)
- ankle (1)
- antegrade puncture (1)
- aorta (1)
- arthrography (1)
- biliäre Stenose (1)
- biliäre Striktur (1)
- biomedical engineering (1)
- bone (1)
- bone remineralization (1)
- bone replacement material (1)
- breast cancer imaging (1)
- cancellous bone (1)
- cancer imaging (1)
- carcinoma (1)
- cardiovascular MRI (1)
- chiasma antebrachii (1)
- chondral defect (1)
- cine loop (1)
- computational neural networks (1)
- computed tomography angiography (CTA) (1)
- continuous extracorporeal femoral perfusion model (1)
- conventional imaging (1)
- convolution kernel (1)
- correlation (1)
- cross-sectional imaging (1)
- digital subtraction angiography (DSA) (1)
- drillability (1)
- dual-energy CT (1)
- dual-source CT (1)
- electrical and electronic engineering (1)
- endovaskuläre Therapie (1)
- experimental models of disease (1)
- femoral access (1)
- femoral arteries (1)
- flexor digitorum superficialis (1)
- flexor tendon (1)
- fluoroscopy (1)
- fracture (1)
- gadoliniumhaltiges Kontrastmittel (1)
- giant cell arteritis (1)
- heart (1)
- high tibial osteotomy (1)
- imaging (1)
- intravascular ultrasound (IVUS) (1)
- intravenös (1)
- knee (1)
- large animal models (1)
- late enhancement (1)
- late gadolinium-enhancement (1)
- lymphatic metastasis (1)
- machine learning (1)
- magnesium phosphate cement (1)
- maligne untere Einflussstauung (1)
- multiple myeloma (1)
- musculoskeletal system (1)
- neoadjuvant therapies (1)
- neuronavigation (1)
- non-small-cell lung (1)
- novel human cadaveric perfusion model (1)
- obesity (1)
- ophthalmic artery (1)
- optic nerve (1)
- orbit (1)
- organoid (1)
- osteoarthritis (1)
- pedicle screws (1)
- perfusion map (1)
- photon-counting (1)
- photon-counting CT (1)
- photon-counting computed tomography (CT) (1)
- photon-counting-detector CT (1)
- phytic acid (1)
- pig (1)
- positive nodal status (1)
- post-neoadjuvant therapies (1)
- preclinical research (1)
- pulmonary embolism (1)
- radiation dosage (1)
- radiation dose (1)
- radiation dose reduction (1)
- radiology (1)
- radiomics (1)
- real-time imaging (1)
- reproducibility of results (1)
- research infrastructure (1)
- small pixel effect (1)
- small-cell lung (1)
- spectral CT (1)
- spectral imaging (1)
- spine (1)
- spiral trajectory (1)
- three-dimensional imaging (1)
- tomography (1)
- traumatisch (1)
- ultrahigh resolution (1)
- ultrahigh-field MRI (1)
- vertebral pedicles (1)
- virtual non-contrast (1)
- whole-body imaging (1)
Institut
- Institut für diagnostische und interventionelle Radiologie (Institut für Röntgendiagnostik) (25) (entfernen)
Objectives
Open-access cancer imaging datasets have become integral for evaluating novel AI approaches in radiology. However, their use in quantitative analysis with radiomics features presents unique challenges, such as incomplete documentation, low visibility, non-uniform data formats, data inhomogeneity, and complex preprocessing. These issues may cause problems with reproducibility and standardization in radiomics studies.
Methods
We systematically reviewed imaging datasets with public copyright licenses, published up to March 2023 across four large online cancer imaging archives. We included only datasets with tomographic images (CT, MRI, or PET), segmentations, and clinical annotations, specifically identifying those suitable for radiomics research. Reproducible preprocessing and feature extraction were performed for each dataset to enable their easy reuse.
Results
We discovered 29 datasets with corresponding segmentations and labels in the form of health outcomes, tumor pathology, staging, imaging-based scores, genetic markers, or repeated imaging. We compiled a repository encompassing 10,354 patients and 49,515 scans. Of the 29 datasets, 15 were licensed under Creative Commons licenses, allowing both non-commercial and commercial usage and redistribution, while others featured custom or restricted licenses. Studies spanned from the early 1990s to 2021, with the majority concluding after 2013. Seven different formats were used for the imaging data. Preprocessing and feature extraction were successfully performed for each dataset.
Conclusion
RadiomicsHub is a comprehensive public repository with radiomics features derived from a systematic review of public cancer imaging datasets. By converting all datasets to a standardized format and ensuring reproducible and traceable processing, RadiomicsHub addresses key reproducibility and standardization challenges in radiomics.
Critical relevance statement
This study critically addresses the challenges associated with locating, preprocessing, and extracting quantitative features from open-access datasets, to facilitate more robust and reliable evaluations of radiomics models.
Key points
- Through a systematic review, we identified 29 cancer imaging datasets suitable for radiomics research.
- A public repository with collection overview and radiomics features, encompassing 10,354 patients and 49,515 scans, was compiled.
- Most datasets can be shared, used, and built upon freely under a Creative Commons license.
- All 29 identified datasets have been converted into a common format to enable reproducible radiomics feature extraction.
Minimally invasive endovascular interventions have become an important tool for the treatment of cardiovascular diseases such as ischemic heart disease, peripheral artery disease, and stroke. X-ray fluoroscopy and digital subtraction angiography are used to precisely guide these procedures, but they are associated with radiation exposure for patients and clinical staff. Magnetic Particle Imaging (MPI) is an emerging imaging technology using time-varying magnetic fields combined with magnetic nanoparticle tracers for fast and highly sensitive imaging. In recent years, basic experiments have shown that MPI has great potential for cardiovascular applications. However, commercially available MPI scanners were too large and expensive and had a small field of view (FOV) designed for rodents, which limited further translational research. The first human-sized MPI scanner designed specifically for brain imaging showed promising results but had limitations in gradient strength, acquisition time and portability. Here, we present a portable interventional MPI (iMPI) system dedicated for real-time endovascular interventions free of ionizing radiation. It uses a novel field generator approach with a very large FOV and an application-oriented open design enabling hybrid approaches with conventional X-ray-based angiography. The feasibility of a real-time iMPI-guided percutaneous transluminal angioplasty (PTA) is shown in a realistic dynamic human-sized leg model.
Das Ziel der Studie bestand darin, die Durchführbarkeit, Effizienz und Sicherheit der perkutanen transluminalen Zangenbiopsie bei biliären Strikturen nach „Cross and push“ Technik am Universitätsklinikum Würzburg zu evaluieren.
Die Gewinnung der Zangenbiopsie bei den insgesamt 17 Patienten erfolgte an den baugleichen Angiographieeinheiten der interventionellen radiologischen Abteilung.
Die Schaffung des biliären Zugangsweges wurde per Feinnadelpunktion oder über einen bereits einliegenden Drainagekatheter durchgeführt. Nach Vordirigieren einer armierten Führungsschleuse bis unmittelbar vor die Stenose wurden die Gewebeproben mittels Biopsiezange entnommen. Auf diese Weise sollte ein Abknicken des Instrumentariums durch die fibrotische Beschaffenheit der Striktur vermieden und so die Gewebegewinnung erleichtert werden. Im Anschluss folgten die PTCD-Neuanlage bzw. der Katheterwechsel sowie die histologische Auswertung der Zangenbiopsie. Die Dokumentation umfasste den technischen Erfolg, die Biopsieanzahl sowie -größe, die Untersuchungs- und die Fluoroskopiezeit, das Dosisflächenprodukt, die Repräsentativität der Biopsien und die Komplikationsrate.
Die technische Erfolgsrate lag bei 100 %. Bei zehn Patienten und Patientinnen (58,8 %) wurde in der Zangenbiopsie ein Malignom bestätigt. Bei sieben Betroffenen (41,2 %) ergab die Zangenbiopsie einen benignen Befund, wobei dieser in zwei Fällen (11,8 %) falsch negativ war. Die Kriterien der Repräsentativität wurden bei elf Patienten und Patientinnen (64,7 %) erfüllt. Die mediane Biopsiezahl betrug 4 und die Biopsiegröße war 1 - 2 mm. Bei einer medianen Untersuchungszeit von 24 min bemaß sich die mediane Fluoroskopiezeit auf 7,6 min. Das mediane DAP für alle Interventionen betrug 2.593 μGym2. Zu einer Majorkomplikation kam es bei einer Patientin (5,9 %).
Bei der perkutanen transluminalen Zangenbiopsie nach „Cross and push“ Technik handelt es sich um ein effektives und sicheres Verfahren zur histologischen Beurteilung biliärer Strikturen mit einer moderaten Komplikationsrate. Die Anwendung sollte besonders bei proximalen Strikturen in Betracht gezogen werden, wenn die ERCP oder alternative Verfahren nicht erfolgreich waren und parallel eine Entlastung der Cholestase via PTCD intendiert oder bereits vorhanden ist. Zur weiteren Evaluation der Methode ist zukünftig die Betrachtung eines größeren Patientenkollektives nötig. Um das Verfahren mit Alternativmethoden vergleichen zu können, bedarf es randomisierter kontrollierter Studien.
Ultra-high field cardiac MRI in large animals and humans for translational cardiovascular research
(2023)
A key step in translational cardiovascular research is the use of large animal models to better understand normal and abnormal physiology, to test drugs or interventions, or to perform studies which would be considered unethical in human subjects. Ultrahigh field magnetic resonance imaging (UHF-MRI) at 7 T field strength is becoming increasingly available for imaging of the heart and, when compared to clinically established field strengths, promises better image quality and image information content, more precise functional analysis, potentially new image contrasts, and as all in-vivo imaging techniques, a reduction of the number of animals per study because of the possibility to scan every animal repeatedly. We present here a solution to the dual use problem of whole-body UHF-MRI systems, which are typically installed in clinical environments, to both UHF-MRI in large animals and humans. Moreover, we provide evidence that in such a research infrastructure UHF-MRI, and ideally combined with a standard small-bore UHF-MRI system, can contribute to a variety of spatial scales in translational cardiovascular research: from cardiac organoids, Zebra fish and rodent hearts to large animal models such as pigs and humans. We present pilot data from serial CINE, late gadolinium enhancement, and susceptibility weighted UHF-MRI in a myocardial infarction model over eight weeks. In 14 pigs which were delivered from a breeding facility in a national SARS-CoV-2 hotspot, we found no infection in the incoming pigs. Human scanning using CINE and phase contrast flow measurements provided good image quality of the left and right ventricle. Agreement of functional analysis between CINE and phase contrast MRI was excellent. MRI in arrested hearts or excised vascular tissue for MRI-based histologic imaging, structural imaging of myofiber and vascular smooth muscle cell architecture using high-resolution diffusion tensor imaging, and UHF-MRI for monitoring free radicals as a surrogate for MRI of reactive oxygen species in studies of oxidative stress are demonstrated. We conclude that UHF-MRI has the potential to become an important precision imaging modality in translational cardiovascular research.
This work deals with the acceleration of cardiovascular MRI for the assessment
of functional information in steady-state contrast and for viability assessment
during the inversion recovery of the magnetization. Two approaches
are introduced and discussed in detail. MOCO-MAP uses an exponential
model to recover dynamic image data, IR-CRISPI, with its low-rank plus
sparse reconstruction, is related to compressed sensing.
MOCO-MAP is a successor to model-based acceleration of parametermapping
(MAP) for the application in the myocardial region. To this end, it
was augmented with a motion correction (MOCO) step to allow exponential
fitting the signal of a still object in temporal direction. Iteratively, this
introduction of prior physical knowledge together with the enforcement of
consistency with the measured data can be used to reconstruct an image
series from distinctly shorter sampling time than the standard exam (< 3 s
opposed to about 10 s). Results show feasibility of the method as well as
detectability of delayed enhancement in the myocardium, but also significant
discrepancies when imaging cardiac function and artifacts caused already by
minor inaccuracy of the motion correction.
IR-CRISPI was developed from CRISPI, which is a real-time protocol
specifically designed for functional evaluation of image data in steady-state
contrast. With a reconstruction based on the separate calculation of low-rank
and sparse part, it employs a softer constraint than the strict exponential
model, which was possible due to sufficient temporal sampling density via
spiral acquisition. The low-rank plus sparse reconstruction is fit for the use on
dynamic and on inversion recovery data. Thus, motion correction is rendered
unnecessary with it.
IR-CRISPI was equipped with noise suppression via spatial wavelet filtering.
A study comprising 10 patients with cardiac disease show medical
applicability. A comparison with performed traditional reference exams offer
insight into diagnostic benefits. Especially regarding patients with difficulty
to hold their breath, the real-time manner of the IR-CRISPI acquisition provides
a valuable alternative and an increase in robustness.
In conclusion, especially with IR-CRISPI in free breathing, a major acceleration
of the cardiovascular MR exam could be realized. In an acquisition
of less than 100 s, it not only includes the information of two traditional
protocols (cine and LGE), which take up more than 9.6 min, but also allows
adjustment of TI in retrospect and yields lower artifact level with similar
image quality.
Photon-counting detector (PCD) CT allows for ultra-high-resolution (UHR) examinations of the shoulder without requiring an additional post-patient comb filter to narrow the detector aperture. This study was designed to compare the PCD performance with a high-end energy-integrating detector (EID) CT. Sixteen cadaveric shoulders were examined with both scanners using dose-matched 120 kVp acquisition protocols (low-dose/full-dose: CTDI\(_{vol}\) = 5.0/10.0 mGy). Specimens were scanned in UHR mode with the PCD-CT, whereas EID-CT examinations were conducted in accordance with the clinical standard as “non-UHR”. Reconstruction of EID data employed the sharpest kernel available for standard-resolution scans (ρ\(_{50}\) = 12.3 lp/cm), while PCD data were reconstructed with both a comparable kernel (11.8 lp/cm) and a sharper dedicated bone kernel (16.5 lp/cm). Six radiologists with 2–9 years of experience in musculoskeletal imaging rated image quality subjectively. Interrater agreement was analyzed by calculation of the intraclass correlation coefficient in a two-way random effects model. Quantitative analyses comprised noise recording and calculating signal-to-noise ratios based on attenuation measurements in bone and soft tissue. Subjective image quality was higher in UHR-PCD-CT than in EID-CT and non-UHR-PCD-CT datasets (all p < 0.001). While low-dose UHR-PCD-CT was considered superior to full-dose non-UHR studies on either scanner (all p < 0.001), ratings of low-dose non-UHR-PCD-CT and full-dose EID-CT examinations did not differ (p > 0.99). Interrater reliability was moderate, indicated by a single measures intraclass correlation coefficient of 0.66 (95% confidence interval: 0.58–0.73; p < 0.001). Image noise was lowest and signal-to-noise ratios were highest in non-UHR-PCD-CT reconstructions at either dose level (p < 0.001). This investigation demonstrates that superior depiction of trabecular microstructure and considerable denoising can be realized without additional radiation dose by employing a PCD for shoulder CT imaging. Allowing for UHR scans without dose penalty, PCD-CT appears as a promising alternative to EID-CT for shoulder trauma assessment in clinical routine.
In this study, the impact of reconstruction sharpness on the visualization of the appendicular skeleton in ultrahigh-resolution (UHR) photon-counting detector (PCD) CT was investigated. Sixteen cadaveric extremities (eight fractured) were examined with a standardized 120 kVp scan protocol (CTDI\(_{vol}\) 10 mGy). Images were reconstructed with the sharpest non-UHR kernel (Br76) and all available UHR kernels (Br80 to Br96). Seven radiologists evaluated image quality and fracture assessability. Interrater agreement was assessed with the intraclass correlation coefficient. For quantitative comparisons, signal-to-noise-ratios (SNRs) were calculated. Subjective image quality was best for Br84 (median 1, interquartile range 1–3; p ≤ 0.003). Regarding fracture assessability, no significant difference was ascertained between Br76, Br80 and Br84 (p > 0.999), with inferior ratings for all sharper kernels (p < 0.001). Interrater agreement for image quality (0.795, 0.732–0.848; p < 0.001) and fracture assessability (0.880; 0.842–0.911; p < 0.001) was good. SNR was highest for Br76 (3.4, 3.0–3.9) with no significant difference to Br80 and Br84 (p > 0.999). Br76 and Br80 produced higher SNRs than all kernels sharper than Br84 (p ≤ 0.026). In conclusion, PCD-CT reconstructions with a moderate UHR kernel offer superior image quality for visualizing the appendicular skeleton. Fracture assessability benefits from sharp non-UHR and moderate UHR kernels, while ultra-sharp reconstructions incur augmented image noise.
Ultra-high-resolution photon-counting detector CT arthrography of the ankle: a feasibility study
(2023)
This study was designed to investigate the image quality of ultra-high-resolution ankle arthrography employing a photon-counting detector CT. Bilateral arthrograms were acquired in four cadaveric specimens with full-dose (10 mGy) and low-dose (3 mGy) scan protocols. Three convolution kernels with different spatial frequencies were utilized for image reconstruction (ρ\(_{50}\); Br98: 39.0, Br84: 22.6, Br76: 16.5 lp/cm). Seven radiologists subjectively assessed the image quality regarding the depiction of bone, hyaline cartilage, and ligaments. An additional quantitative assessment comprised the measurement of noise and the computation of contrast-to-noise ratios (CNR). While an optimal depiction of bone tissue was achieved with the ultra-sharp Br98 kernel (S ≤ 0.043), the visualization of cartilage improved with lower modulation transfer functions at each dose level (p ≤ 0.014). The interrater reliability ranged from good to excellent for all assessed tissues (intraclass correlation coefficient ≥ 0.805). The noise levels in subcutaneous fat decreased with reduced spatial frequency (p < 0.001). Notably, the low-dose Br76 matched the CNR of the full-dose Br84 (p 0.999) and superseded Br98 (p < 0.001) in all tissues. Based on the reported results, a photon-counting detector CT arthrography of the ankle with an ultra-high-resolution collimation offers stellar image quality and tissue assessability, improving the evaluation of miniscule anatomical structures. While bone depiction was superior in combination with an ultra-sharp convolution kernel, soft tissue evaluation benefited from employing a lower spatial frequency.
Die interventionelle Radiologie hat sich in den letzten Jahrzehnten zunehmend auf palliativmedizinische sowie onkologische Bereiche ausgeweitet und kann durch minimal-invasive Therapieoptionen gerade in vulnerablen Patientenkollektiven attraktive Behandlungsmöglichkeiten zur Verfügung stellen. Die tumorbedingte untere Einflussstauung ist ein seltenes Krankheitsbild und stellt eine schwere symptomatische Komplikation einer malignen Grunderkrankung dar. Dabei kommt es im Rahmen dieser Grunderkrankung durch die Primärtumormasse oder Metastasen zu extrinsischer Kompression der Vena cava inferior (VCI), Gefäßinvasion oder Thrombusbildung.
Ziel der Dissertationsarbeit ist es, den technischen und klinischen Erfolg der Sinus-XL ® Stentimplantation in die Vena cava inferior bei einer tumorbedingten unteren Einflussstauung zu untersuchen. Als technischer Erfolg wurde dabei die problemlose Stentimplantation mit anschließender Aufhebung der VCI-Stenose/Okklusion und Revaskularisation der VCI definiert. Bezüglich des klinischen Erfolges wurde der Frage nachgegangen, inwieweit die Stentimplantation die typischen Symptome einer unteren Einflussstauung (Ödeme der unteren Extremität, Aszites und Anasarka) lindern und bestenfalls eliminieren kann. In der vorliegenden Arbeit sind dazu retrospektiv die Daten von insgesamt 21 Patienten (11 Frauen, 10 Männer) mit einem medianen Alter von 61 Jahren (19-92 Jahre), die zwischen Oktober 2010 und Januar 2021 aufgrund einer tumorbedingten unteren Einflussstauung mit einem Sinus-XL ® Stent endovaskulär versorgt wurden, ausgewertet worden. Zur Quantifizierung der klinischen Symptomatik wurde für das jeweilige Symptom ein Scoring-System entwickelt bzw. modifiziert.
Der technische Erfolg belief sich auf 100% (21/21). Postinterventionell konnte zudem eine signifikante Reduktion des transstenotischen Druckgradienten (p = 0,008) und eine signifikante Aufweitung des Stenosendiameters (p < 0,001) erreicht werden. Die primäre und primär-assistierte Stentoffenheit betrug 92,9 % (13/14) und 100% (14/14), die anatomische Stentoffenheit (< 50% Restenose) belief sich auf 53,3 % (8/15). Die Reinterventionsrate lag bei 4,8 % (1/21). Schwerwiegende Komplikationen traten nicht auf. Der klinische Erfolg bezüglich der Ödeme der unteren Extremität belief sich auf 82,4 % (14/17), 93,8 % (15/16) sowie auf 85,7 % (18/21) und zeigte in allen betrachteten Zeitintervallen eine signifikante Scorewertreduktion (p < 0,001). Das klinische Outcome bezüglich der Ödeme war bei kürzeren Stenosen/Obstruktionen signifikant besser (p = 0,025). Bezüglich einer intrahepatischen Segmentbeteiligung, der transstenotischen Druckgradienten, der absoluten Gradientenreduktion sowie der Überlebenszeit nach der Intervention zeigten sich hingegen keine als klinisch relevant einzustufende Ergebnisse. Ein eindeutiger Effekt der Intervention auf die Symptome Anasarka und Aszites konnte nicht nachgewiesen werden. Diesbezüglich zeigten sich klinische Erfolgsraten von 42,9 % (6/14) und 5,3 % (1/19). Im postinterventionellen Verlauf konnten außerdem signifikante Reduktionen der präinterventionellen Harnstoffwerte sowie des Körpergewichtes der Patienten verzeichnet werden.
Zusammenfassend zeigt die vorliegende Arbeit, dass die Sinus-XL ® Stentimplantation geeignet ist, eine tumorbedingte Vena cava inferior-Stenose/Obstruktion aufzuheben und eine Revaskularisation der VCI zu erreichen. Die klinischen Symptome einer unteren Einflussstauung – insbesondere bezogen auf die Ödeme der unteren Extremität und mit Einschränkungen bezogen auf die Symptome Aszites und Anasarka – können ebenfalls durch die Stentimplantation gelindert und teilweise sogar langanhaltend eliminiert werden. Die Sinus-XL ® Stentimplantation sollte daher stets als Therapieoption bei tumorbedingten unteren Einflussstauungen in Erwägung gezogen werden. Nicht zuletzt stellt die Stentimplantation auch eine sichere und komplikationsarme Intervention dar. Weitere Studien, bestenfalls multizentrische Studien, sind jedoch notwendig, um die dargestellten Ergebnisse weiter zu untermauern.
Objectives: Positron emission tomography (PET) is currently considered the non-invasive reference standard for lymph node (N-)staging in lung cancer. However, not all patients can undergo this diagnostic procedure due to high costs, limited availability, and additional radiation exposure. The purpose of this study was to predict the PET result from traditional contrast-enhanced computed tomography (CT) and to test different feature extraction strategies. Methods: In this study, 100 lung cancer patients underwent a contrast-enhanced \(^{18}\)F-fluorodeoxyglucose (FDG) PET/CT scan between August 2012 and December 2019. We trained machine learning models to predict FDG uptake in the subsequent PET scan. Model inputs were composed of (i) traditional “hand-crafted” radiomics features from the segmented lymph nodes, (ii) deep features derived from a pretrained EfficientNet-CNN, and (iii) a hybrid approach combining (i) and (ii). Results: In total, 2734 lymph nodes [555 (20.3%) PET-positive] from 100 patients [49% female; mean age 65, SD: 14] with lung cancer (60% adenocarcinoma, 21% plate epithelial carcinoma, 8% small-cell lung cancer) were included in this study. The area under the receiver operating characteristic curve (AUC) ranged from 0.79 to 0.87, and the scaled Brier score (SBS) ranged from 16 to 36%. The random forest model (iii) yielded the best results [AUC 0.871 (0.865–0.878), SBS 35.8 (34.2–37.2)] and had significantly higher model performance than both approaches alone (AUC: p < 0.001, z = 8.8 and z = 22.4; SBS: p < 0.001, z = 11.4 and z = 26.6, against (i) and (ii), respectively). Conclusion: Both traditional radiomics features and transfer-learning deep radiomics features provide relevant and complementary information for non-invasive N-staging in lung cancer.