ECP 2017 - Abstract Book

Conclusion:

Such study concluded that DIA of the snapshots ob-

tained from the virtual slides is still the most accurate method for

evaluation of Ki67 in the cases of cancer breast.

PS-14-019

Impact of a laboratory information system - integrated speech rec-

ognition system as part of digitisation in pathology departments

A. Nieto Olivares

, E. Alcaraz-Mateos, F. García-Molina, A. Chaves-

Benito

Hospital Morales Meseguer, Pathology, Murcia, Spain

Objective:

Digitization includes not only the scanning process, but also

the traceability of tissue samples and report management. In our opinion,

this last point has been insufficiently studied. Currently, in our center,

resource limitations, mainly human, have resulted in the task of pathology

report editing to be assigned to the pathologist, with the consequent work

overload related thereof. With the aim of achieving time optimization, a

speech recognition system (SRS) was implemented.

Method:

Two pathologists from our department, specialized in gas-

trointestinal (GI) pathology, were involved in the implementation of a

Laboratory Information System (LIS)

–

integrated SRS (INVOX

Medical Dictation, Vocali). With a preliminary training period of

2 weeks, a total of 60 GI pathology cases were randomly selected,

with the intention of comparing the reporting time for both typing

and dictation. Several abbreviations and shortcuts were established

when using the SRS, including SNOMED CT automatic coding.

Results:

The overall reporting time for the 60 cases was 95

′

″

′

″

/case) with the typing methodology, while for the SRS it was 42

′

″

(43

″

/case).This meant a reduction in reporting time of 55.2 %.

In addition, there was a standardization of the pathology reports,

both for structured texts and those edited as continuous text.

Conclusion:

A SRS can help pathologists optimize time in signing

out cases, making them more efficient and cost-effective.

Abbreviations and shortcuts allow reports to be more standardized,

while avoiding typing errors. At the same time, this methodology

implies a lower physical demand, as it is more ergonomic and re-

duces work-related musculoskeletal disorders. Implementation of

SRS should be considered in the context of going fully digital.

PS-14-021

A model for simulation of glands on virtual slides

M. H. Mandrela

, S. Lohmann, I. Klempert, S. Elezkurtaj, P. Hufnagl

Charité Universitätsmedizin, Abt. Pathologie, Berlin, Germany

Objective:

Big Data revolutionized machine learning and facilitated new

forms of automation. It is possible to use deep learning to perform object

segmentation of images given large enough training sets. The automatic

segmentation and analysis of glands in HE stained tissue slides could

facilitate the detection of all types of glands and support quantification.

But training data is scarce, because of the time consuming and error prone

process of labeling data by pathologists.

Method:

We aim to bridge the gap in training data by synthesizing

simulated HE-slides (phantoms) of prostate and colon gland tissue.

A complex model based on distances, distributions, geometrical

shapes and statistical processes was created. Texture was simulated

using a neural network.

Results:

This model randomly generates phantoms based on the given

parameters which allows the creation of images of different tissue types

with varying gland forms and types.

Conclusion:

A model for the description of glands was developed which

can be used to supplement training data for machine learning applications

or as a benchmark for algorithm development. The model was evaluated

using the generated phantoms to train a neural model for image

segmentation, by statistical comparison to real images and it was exam-

ined by pathologists.

PS-15-001

Clinicopathological study of amyloidosis in renal biopsy

H. Oliveira Coelho

, P. Pereira Campos, G. Martins-Coelho, M. J. Brito,

F. Carvalho

Hospital Garcia de Orta, Anatomia Patológica, Almada, Portugal

Objective:

This study characterizes the clinicopathological features of

patients with renal amyloidosis.

Method:

We reviewed slides and clinical data of all amyloidosis

cases (

= 62) in renal biopsies between 2011 and 2016. Amyloid

characterization was carried out by immunohistochemistry (A pro-

tein and TTR) and direct immunofluorescence (light chains k and

). Amyloid deposition was classified by Congo red stain accord-

ing to its quantity (none, mild and high) and localisation (glomer-

ular, vascular and tubulointerstitial).

Results:

Mean age was 60; 31 (50 %) were women. The most

frequent clinical change was nephrotic proteinuria/syndrome [43

patients (69 %)]. Six patients (10 %) were HIV+. Amyloidosis

breakdown by type was: AA

—

27 cases (44 %); AL

—

24 cases

(38 %) [k chain

—

2 (3 %);

chain

—

22 (35 %)]; ATTR cases

—

(10 %); and undetermined

—

5 cases (8 %). All cases had vascular

deposits, 39 (63 %) of which were high quantity; 97 % cases had

glomerular deposits [38 (61 %) high quantity], and 43 % had

tubulointerstitial deposits [9 (15 %) high quantity]. Amyloid quan-

tity was similar in the different types (

= 0,33).

Conclusion:

Amyloidosis is mainly a disease of older individuals.

AA and AL were the most common forms. Amyloid was mainly

deposited in the vessels and glomeruli. There was no association

between the quantity and the type of amyloid.

PS-15-003

Findings in percutaneous renal biopsy: 12-year experience

O. Cambero

, L. Alegre, L. Lozano, A. Puente, Á. Castaño, C. Ruiz de

Valbuena

H. U. de Fuenlabrada, Dept. of Pathology, Spain

Objective:

To present our experience and outcome of percutaneous

renal biopsies for the diagnosis of nephrological and systemic

diseases.

Method:

Consecutive 242 percutaneous renal biopsies, taken from 232

patients and performed over a 12-year period were retrieved from our

files. Renal biopsies were studied by light, immunofluorescence and elec-

tron microscopy.

Results:

The mean age of the population was 48,8 years (range 14

–

89; 95 females and 137 males). Clinical indications which lead to

renal biopsy were: acute renal failure (ARF, 26,4 %), nephrotic

syndrome (NS, 19,8 %), chronic renal failure (CRF, 19 %), protein-

uria (18,6 %) and proteinuria-haematuria (13,6 %). Pathological

diagnosis were: IgA nephropathy (19,8 %), lupus nephritis (9 %),

glomerulosclerosis (7,8 %), diabetic nephropathy (7 %), membra-

nous glomerulopathy (7,9 %), minimal change disease (7 %), vas-

cular diseases (5,4 %), tubulointerstitial nephritis (5,4 %), vasculitis

( 3 , 7 % ) , m e s a n g i a l g l o m e r u l o n e p h r i t i s ( 3 , 3 % ) ,

membranoproliferative glomerulonephritis (3,7 %), amyloidosis

Tuesday, 5 September 2017, 09:30

–

10:30, Hall 3

PS-15 Nephropathology

Virchows Arch

(

2017

)

471

(

Suppl 1

–

S352

S207