Performance of four dual-energy CT platforms for abdominal imaging: a task-based image quality assessment based on phantom data.

OBJECTIVES: To compare the spectral performance of dual-energy CT (DECT) platforms using task-based image quality assessment based on phantom data. MATERIALS AND METHODS: Two CT phantoms were scanned on four DECT platforms: fast kV-switching CT (KVSCT), split filter CT (SFCT), dual-source CT (DSCT), and dual-layer CT (DLCT). Acquisitions on each phantom were performed using classical parameters of abdomen-pelvic examination and a CTDIvol at 10 mGy. Noise power spectrum (NPS) and task-based transfer function (TTF) were evaluated from 40 to 140 keV of virtual monoenergetic images. A detectability index (d') was computed to model the detection task of two contrast-enhanced lesions as function of keV. RESULTS: The noise magnitude decreased from 40 to 70 keV for all DECT platforms, and the highest noise magnitude values were found for KVSCT and SFCT and the lowest for DSCT and DLCT. The average NPS spatial frequency shifted towards lower frequencies as the energy level increased for all DECT platforms, smoothing the image texture. TTF values decreased with the increase of keV deteriorating the spatial resolution. For both simulated lesions, higher detectability (d' value) was obtained at 40 keV for DLCT, DSCT, and SFCT but at 70 keV for KVSCT. The detectability of both simulated lesions was highest for DLCT and DSCT. CONCLUSION: Highest detectability was found for DLCT for the lowest energy levels. The task-based image quality assessment used for the first time for DECT acquisitions showed the benefit of using low keV for the detection of contrast-enhanced lesions. KEY POINTS: • Detectability of both simulated contrast-enhanced lesions was higher for dual-layer CT for the lowest energy levels. • The image noise increased and the image texture changed for the lowest energy levels. • The detectability of both simulated contrast-enhanced lesions was highest at 40 keV for all dual-energy CT platforms except for fast kV-switching platform.

Impact of ultra-low dose CT acquisition on semi-automated RECIST tool in the evaluation of malignant focal liver lesions.

PURPOSE: To compare the evaluation of malignant focal liver lesions (FLLs) using a semi-automated RECIST tool with a standard and an ultra-low dose (ULD) computed tomography (CT) protocol. MATERIALS AND METHODS: Thirty-four patients with malignant FLLs underwent two abdominal-pelvic CT examinations one using a standard protocol and one using an ULD protocol. There were 23 men and 11 women with a mean age 64.3±14.4 (SD) years (range: 22-91 years). Dosimetric indicators were recorded, and effective dose was calculated for both examinations. Mean malignant FLL attenuation, image noise and contrast-to-noise-ratio (CNR) were compared. The largest malignant FLL per patient was evaluated using the semi-automated RECIST tool to determine longest axis length, longest orthogonal axis length, volume and World Health Organisation area. RESULTS: Dosimetric values were significantly reduced by -56% with ULD compared to standard protocol. No differences in mean malignant FLL attenuation values were found between the two protocols. Image noise was significantly increased for all locations (P<0.05) with ULD compared to standard protocol, and CNR was significantly reduced (P<0.05). On the 34 malignant FLLs analyzed, six semi-automated shapes non-concordant with radiologist's visual impression were highlighted with the software, including one FLL (1/34; 3%) with standard CT acquisition only, three FLLs (3/34; 9%) with ULD CT acquisition only and two FLLs (2/34; 6%) with both CT acquisitions. After manual editing, the concordance of the values of the studied criteria between both acquisitions was good and no significant difference was reported. CONCLUSION: Semi-automated RECIST tool demonstrates good performances using ULD CT protocol. It could be used in routine clinical practice with a ULD protocol for follow-up studies in patients with known malignant FLL.

Early MRI predictors of disease-free survival in locally advanced rectal cancer from the GRECCAR 4 trial.

BACKGROUND: Tailored neoadjuvant treatment of locally advanced rectal cancer (LARC) may improve outcomes. The aim of this study was to determine early MRI prognostic parameters with which to stratify neoadjuvant treatment in patients with LARC. METHODS: All patients from a prospective, phase II, multicentre randomized study (GRECCAR4; NCT01333709) were included, and underwent rectal MRI before treatment, 4 weeks after induction chemotherapy and after completion of chemoradiotherapy (CRT). Tumour volumetry, MRI tumour regression grade (mrTRG), T and N categories, circumferential resection margin (CRM) status and extramural vascular invasion identified by MRI (mrEMVI) were evaluated. RESULTS: A total of 133 randomized patients were analysed. Median follow-up was 41·4 (95 per cent c.i. 36·6 to 45·2) months. Thirty-one patients (23·3 per cent) developed tumour recurrence. In univariable analysis, mrEMVI at baseline was the only prognostic factor associated with poorer outcome (P = 0·015). After induction chemotherapy, a larger tumour volume on MRI (P = 0·019), tumour volume regression of 60 per cent or less (P = 0·002), involvement of the CRM (P = 0·037), mrEMVI (P = 0·026) and a poor mrTRG (P = 0·023) were associated with poor outcome. After completion of CRT, the absence of complete response on MRI (P = 0·004), mrEMVI (P = 0·038) and a poor mrTRG (P = 0·005) were associated with shorter disease-free survival. A final multivariable model including all significant variables (baseline, after induction, after CRT) revealed that Eastern Cooperative Oncology Group performance status (P = 0·011), sphincter involvement (P = 0·009), mrEMVI at baseline (P = 0·002) and early tumour volume regression of 60 per cent or less after induction (P = 0·007) were associated with relapse. CONCLUSION: Baseline and early post-treatment MRI parameters are associated with prognosis in LARC. Future preoperative treatment should stratify treatment according to baseline mrEMVI status and early tumour volume regression.

ANTECEDENTES: El tratamiento neoadyuvante personalizado del cáncer de recto localmente avanzado (locally advanced rectal cancer, LARC) puede mejorar los resultados. El objetivo de este estudio fue determinar factores pronósticos precoces mediante RMN para estratificar el tratamiento neoadyuvante en pacientes con LARC. MÉTODOS: Todos los pacientes de un eensayo prospectivo de fase II, multicéntrico y aleatorizado (GRECCAR4-NCT01333709) se incluyeron en este estudio y se les realizó una RMN antes del tratamiento, 4 semanas después de la quimioterapia de inducción y después de completar la quimiorradioterapia (chemoradiation, CRT). Se evaluó la volumetría tumoral, el grado de regresión tumoral mediante RMN (MRI Tumor Regression Grade, mrTRG), la estadificación T, la estadificación N, el estado del margen de resección circunferencial (circumferential resection margin, CRM) y la presencia de invasión extramural vascular en la RMN (extramural vascular invasion, mrEMVI). RESULTADOS: Se analizaron 133 pacientes aleatorizados. La mediana de seguimiento fue de 41,4 meses (i.c. del 95%: 36,6-45,2). En 31 pacientes (23%) se diagnosticó una recidiva. En el análisis univariado de la situación basal, mrEMVI fue el único factor pronóstico asociado con un peor resultado (P = 0,0152). Después de la quimioterapia de inducción, un volumen tumoral más alto en la RMN (P = 0,019), una regresión del volumen tumoral ≤ 60% (P = 0,002), la afectación del CRM (P = 0,037), mrEMVI (P = 0,026) y un grado escaso mrTRG (P = 0,023) se asociaron con un mal resultado. Después de completar la CRT, la ausencia de respuesta completa en la RMN (P = 0,004), la presencia de mrEMVI (P = 0,04) y una insuficiente mrTRG (P = 0,005) se asociaron con una supervivencia libre de enfermedad más corta. En el modelo multivariable final en el que se incluyeron todas las variables significativas (basales, postinducción, post-CRT), el estado de ECOG (P = 0,011), la afectación esfinteriana (P = 0,009), la presencia de EMVI al inicio (P = 0,002) y una regresión precoz del volumen tumoral ≤ 60% después de la inducción (P = 0,007) se asociaron con una recidiva. CONCLUSIÓN: Los parámetros basales y post-tratamiento precoces de la RMN se asocian con el pronóstico en el LARC. La estrategia terapéutica preoperatoria futura deberá estratificar el tratamiento de acuerdo con la presencia de EMVI al inicio y la regresión precoz del volumen tumoral.

Prospective medical analysis of radiation therapist image repositioning during image-guided radiotherapy.

PURPOSE: Radiation oncologists are responsible for deciding which day-to-day variations are acceptable or not in the treatment setup. However, properly qualified and trained radiation therapists might be capable to perform image registration. We evaluated in our centre the capability and accuracy of radiation therapists to validate positioning images in a prospective study. METHODS AND PATIENTS: A total of 84 patients treated for prostate, head and neck, lung or breast cancer was prospectively and randomly included from July 2011 to July 2013 in radiotherapy unit of our institution. For each patient, three positioning images were randomly analysed. Two radiation oncologists analysed all positioning images and shifts decided by the radiation therapists in an independent and blinded way. The radiation oncologists had to decide whether to validate or not this shift and give a corresponding additional shift, if any. A theoretical disagreement rate less than 5% between radiation therapists and radiation oncologists was planned. RESULTS: A total of 240 images were analysed (head and neck: 15.0%; prostate: 14.2%; breast: 55.0%; lung: 15.8%). The global disagreement between radiation oncologists and radiation therapists for all the images analysed was 2.5% 95% confidence interval (95% CI) [1.0-5.0], corresponding to six images out of 240. A 100% agreement was reached for prostate and lung images, a 97.2% agreement for head and neck images and a 96.2% agreement for breast images. CONCLUSIONS: The radiation therapist validation for repositioning images seemed accurate for image-guided radiotherapy in our institution. Periodic evaluation and in-house training are warranted when routine delegation of image registration to radiation therapists is considered.

MRI-based score to predict surgical difficulty in patients with rectal cancer.

BACKGROUND: Rectal cancer surgery is technically challenging and depends on many factors. This study evaluated the ability of clinical and anatomical factors to predict surgical difficulty in total mesorectal excision. METHODS: Consecutive patients who underwent total mesorectal excision for locally advanced rectal cancer in a laparoscopic, robotic or open procedure after neoadjuvant treatment, between 2005 and 2014, were included in this retrospective study. Preoperative clinical and MRI data were studied to develop a surgical difficulty grade. RESULTS: In total, 164 patients with a median age of 61 (range 26-86) years were considered to be at low risk (143, 87·2 per cent) or high risk (21, 12·8 per cent) of surgical difficulty. In multivariable analysis, BMI at least 30 kg/m2 (P = 0·021), coloanal anastomosis (versus colorectal) (P = 0·034), intertuberous distance less than 10·1 cm (P = 0·041) and mesorectal fat area exceeding 20·7 cm2 (P = 0·051) were associated with greater surgical difficulty. A four-item score (ranging from 0 to 4), with each item (BMI, type of surgery, intertuberous distance and mesorectal fat area) scored 0 (absence) or 1 (presence), is proposed. Patients can be considered at high risk of a difficult or challenging operation if they have a score of 3 or more. CONCLUSION: This simple morphometric score may assist surgical decision-making and comparative study by defining operative difficulty before surgery.